2014 and 2015 were golden years for the pharmaceutical industry because the FDA approved an unusually large number of new drugs: 44 in 2014 (the most since 1996), 51 in 2015 (the most since 1950). The top three companies are always the same (J&J, Glaxo and Novartis), but more than 50% of the new drugs were not developed by "big pharma"; and the share of "biologicals" keeps increasing: 22% in 2013, 35% n 2014 (16 out of 44), 39% in 2015.

The story is a bit more complex. First of all, the cost of genome testing keeps falling because the cost of the machines keeps falling. The gene-sequencing market is dominated by three companies: San Diego-based Illumina (that had acquired the sequencing technology of Solexa in 2007), Silicon Valley-based Applied Biosystems (acquired in 2014 by Thermo Fisher Scientific), and 454 Corporation (founded by Jonathan Rothberg in 1999 in Connecticut and acquired by Roche in 2007. Illumina has about 70% of the market. As their machines get cheaper, the product sold to the customers also get cheaper.

That's the good news: DNA testing is getting cheaper and cheaper. The bad news is that it is not very useful: the results of the DNA test are not "actionable", e.g. it doesn't tell you what you should do to reduce the risk of a disease. In fact, often the price does not include an analysis of the results. Neither 23andMe nor Illumina included the "interpretation" of the data with their cheapest genome testing. Competitors of 23andme have multiplied all over the world, but only a few of these "personal-genomic" startups provide comprehensive reports, reports that a health-care specialist can use to make real predictions. What we need now is personal genomics for "predictive" medicine. Two leaders in "actionable" reports are Boston-base Knome (that was the first startup to introduce a commercial human genome sequencing in 2007, which was acquired by Utah-based Tute Genomics, which was then acquired by St Louis-based PierianDx) and Illumina; but they charge more than $10,000 for this kind of reports. In 2015 Maryland-based Veritas Genetics, which has a research center in Hangzhou, a startup founded in 2014 by George Church (the director of Harvard's Personal Genome Project), announced a package that includes both "sequencing" and "interpretation" of the genome for $1,000. In 2016 Las Vegas-based Sure Genomics, founded in 2014 by people who have no background in biological sciences, announced a way for its customers to do the tests at home with a single saliva test) and receive a more comprehensive DNA test than the one they can get from 23andme. The price is higher than 23andme's package ($2,500) but its report is supposed to be much better.

In 2017 Inova and Veritas Genetics got together and launched MyMap, that can provide advice for diet and lifestyle. Veritas' MyGenome screens for 1,200 conditions and 100 inherited conditions, and Inova's MediMap analyzes 31 genes that influence response to 145 prescription medications.

The next revolution in gene-sequencing will come when we have portable gene-sequencers. That day is very near. In 2012 Oxford Nanopore (founded in 2005 by a professor of Chemical Biology at the University of Oxford, Hagan Bayley) began testing a portable gene sequencer called Minion. It was immediately used by doctors to "read" the genomes of Ebola viruses in Guinea during the epidemics that killed 20,000 people. You can plug the Minion into the USB port of a laptop so it will display the results in real time. It is even better than a "lab-on-a-chip": it is a "lab-on-a-USB-drive". The first release only worked well with short genomes, not with genomes as long and complex as the human genome, but in the near future portable devices like the Minion will come to the market. A biologist can pack it in her backpack and take it with her into the jungle to sequence the DNA of some rare animal; the police can use it to quickly identify unknown organisms that could be biological weapons; NASA can use it to test the surface of Mars for signs of life.

There are many possible applications of genomics. You may want to find out if you have any European ancestors. Or you may want to find out whether you are likely to have Alzheimer. Or what sports you are more likely to succeed in. Boston-based startup Good Start Genetics tells parents if their children are likely to have a serious genetic disease like cystic fibrosis. These are different genomic "apps". When you buy a book on Amazon, Amazon suggests other books. Helix will be able to suggest other "apps" that can give you important information about your DNA. Helix, a 2015 San Francisco-based Illumina spinoff, wants to create the first "app store" for genetic information. The customer of a genetic application will have the choice to share the genetic information delivered by that application with other genetic applications.

Craig Venter founded Human Longevity Inc in San Diego in 2013, a company that has been studied genetic data and already found some correlations between genetic variations and longevity. Ambrosia, a startup in Monterey founded by Jesse Karmazin, has begun experimental transfusions of younger blood to older people because Stanford's scientist Tony Wyss-Coray discovered that rats live longer when given the blood of younger rats. Tony Wyss-Coray's student Saul Villeda, lead author of the 2014 study on young blood transfusions, went on to open his own lab at UCSF. This research was a continuation of studies originally published in 2005 by Thomas Rando's laboratory at Stanford ("Rejuvenation of Aged Progenitor Cells by Exposure to a Young Systemic Environment", 2005).

For me the science of longevity really begins in 1993. In 1993 Cynthia Kenyon at UC San Franisco discovered that partially disabling a gene called Daf-2 can double the life of a worm (to one month instead of two weeks). This became known as the sugar problem because eating sugar basically is equivalent to activating (instead of disabling) Daf-2, and in fact sugar shortens the life of the worm (Kenyon famously warned that "sugar is the new tobacco"). After that experiment many other experiments focused on genes and chemicals of all types that seem to affect the lifespan of animals and plants.

A few years later (1999) Leonard Guarente at the MIT found a gene that increased the lifespan of yeast, SIR2. SIRT1 is the equivalent gene in mammals and the family of these genes, "sirtuin", became known as "the anti-aging gene". Guarente and Cynthia Kenyon founded Elixir Pharmaceuticals in 1999 to make anti-aging products. In 2003 Fritz Muller's team at the University of Fribourg in Switzerland discovered that suppressing an enzyme called TOR (Target of Rapamycin) increased the lifespan of worms. Zelton Dave Sharp at the University of Texas proved that the same is true in mice: give them rapamycin (the inhibitor of TOR) and they live longer lives. In 2007 Guarente's pupil David Sinclair showed that the two substances (sirtuin and rapamycin) target the same "longevity pathway". Biologists started searching for "sirtuin activators" and "TOR inhibitors". There was one obvious TOR inhibitor, rapamycin, but no obvious sirtuin activator. In 2003 Sinclair had proposed resveratrol as sirtuin activator, a substance that is found in red wine, and had founded Sirtris in 2004 to make anti-aging drugs based on resveratrol. In 2008 GlaxoSmithKline bought Sirtris, but now the scientific consensus is that resveratrol doesn't work, especially after a study conducted in 2014 by Richard Semba's team of Johns Hopkins University. On the other hand, rapamycin (known as "rapamune" by pharmacies around the world) definitely works on mice. It was confirmed by another study in 2009 led by Richard Miller of the University of Michigan but conducted by three separate teams in three different universities (the other two project leaders were David Harrison at the Jackson Laboratories in Maine and Randy Strong at the University of Texas in San Antonio)

Soon the enthusiasm spread to Silicon Valley. Aubrey de Grey, a former Artificial Intelligence scientist, published the book "Ending Aging" (2007) and founded the SENS (Strategies for Engineered Negligible Senescence) Research Foundation in 2009 in Mountain View.

These studies are interesting but we should never forget that dying is not really a disease. What happens to everybody is not a disease: it is the norm. Everybody dies, so that is not a disease in the sense that diabetes or malaria are diseases. When we look for a drug to cure malaria, we are looking for a drug to turn malaria victims into normal people. When we look for a drug to make us immortal, we are looking for a drug to turn people into something else, not people anymore.

The search for longevity, and possibly even immortality, has led scientists to a tiny polyp, the hydra. This is the only animal that does not age and therefore does not die of old age. It would be immortal if no predator killed it. What makes the hydra "immortal" is that its stem-cells keep proliferating. In 2012 Thomas Bosch at Kiel University discovered that this property of the hydra is due to the so-called FoxO gene, which all animals have but only in some individuals it works overtime. Scientists have long suspected that this gene is important for longevity because a 2008 study by David Curb's team (mainly Bradley Willcox) at the University of Hawaii showed that this gene seems to be particularly active in centenarians. Some day with a bit of genetic manipulation it may be someday possible for humans to regenerate stem-cells and increase longevity. (Bad news for tall people: in 2014 a study by the same team showed that FoxO3 was "inversely associated with height", i.e. the taller you are the shorter your life expectancy. But good news for tea drinkers: the same team showed that drinking tea helps activate FoxO gene expression, i.e. live longer lives. But, don't panic, all these studies are very preliminary).

Another interesting animal for the study of longevity is the jellyfish called "turritopsis". It is the only known animal that can reverse its life cycle and rejuvinate. This animal is not immortal (it does die) but for brief periods it can get younger, and that's certainly something that many elderly people would like to do.

A chemical called NRF-2 became famous when in 2010 Rochelle Buffenstein at the University of Texas showed that it is a key actor in the aging process while protecting the body from diseases. In 2016 Linda Partridge of University College London lithium prolongs the life of fruit flies because (as known since 1996 thanks to the thesis at the University of Toronto by Jim Woodget's student Vuk Stambolic) it blocks a chemical called GSK-3 (that is suspected of being involved in the aging process) while at the same time stimulating that famous NRF-2.

Emma Teeling at University College Dublin has been studying the Myotis bats, that could be the longest-living mammals relative to size, and she discovered that their telomeres do not shorten with age, unlike what happens with us humans and most animals ("Growing Old, Yet Staying Young - The Role of Telomeres in Bats' Exceptional Longevity", 2018).

In 2016 Manfred Kayser in the Netherlands discovered that a gene called MC1R is responsible for "looking older".

In 1960 Leonard Hayflick discovered cellular senescence: cells age and the immune system should get rid of them when they are "senescent" but as we get older the immune system becomes less and less skilled at this job. The accumulation of these senescent cells with ageing causes chronic disorders that not only shorten our life but make our old age painful. Senolytic drugs (or just "senolytics") are medicines that clear senescent cells out of the body: Dasatinib, Quercetin, Fisetin and Navitoclax were the first ones.

In 2011 Jan VanDeursen's team at the Mayo Clinic found a way to remove senescent cells from mice. These senescent cells are almost literally rotting, secreting a toxin called SASP (Senescence-Associated Secretory Phenotype) that contaminates nearby cells spreads senescence throughout the body. Van Deursen immediately co-founded Unity Biotechnology in San Francisco to commercialize this idea (together with Judith Campisi of the Buck Institute for Research on Aging and serial biotech entrepreneur Nathaniel David). Their investors have included billionaires like Jeff Bezos and Peter Thiel.

In 2018 the IPO for Unity Biotechnology was one of the biggest of the year. It came when Unity had just started the first clinical trials of "senolytics". Senolytics are chemical compounds that eliminate senescent cells in human tissues, i.e. can slow or even reverse the aging process. In 2018 Judith Campisi at the Buck Institute removed senescent cells from the brains of mice, and at the same time Darren Baker's team at the Mayo Clinic did the same to mice with Alzheimer. In both cases the mice seem to get rejuvinated. A different technique was used at the same time (2018) by Paul Robbins and Laura Niedernhofer (the directors of the Institute on the Biology of Aging and Metabolism at the University of Minnesota: they used fisetin to eliminate the senescent cells.

In 2000 Shinichiro Imai of Guarente's team discovered that the action of sirtuins depends on Nicotinamide Adenine Dinucleotide or NAD ("Transcriptional Silencing and Longevity Protein Sir2 is an NAD-dependent Histone Deacetylase", 2000) and this started a whole line of research into NAD as an agent of longevity: NAD basically enhances the work of mitochondria, the cell's organs that control our metabolism, but its production declines with age. This means that the energy levels in the cell decline with age. So Imai had discovered that the beneficial effects of sirtuins really depends on how "energetic" the cell is, and that energy depends on mitochondria, and that depends on NAD. Unfortunately, you cannot administer NAD to people, so most experiments consist in administering the chemicals that "boost" NAD in people like Nicotinamide Riboside (NR) and Nicotinamide MonoNucleotide (NMN). Resveratrol activates SIRT1 alone, whereas these boosters activate all seven sirtuins. It was known that many problems of aging are the result of bad mutations in mitochondrial functions, but it was believed that these mutations cannot be reversed. After David Sinclair of resveratrol fame showed that NMN reversed muscle aging in mice ("Declining NAD+ Induces a Pseudohypoxic State Disrupting Nuclear-Mitochondrial Communication during Aging", 2013) a number of startups were launched to work on drugs that can produce NAD in our bodies: EdenRoc, Liberty Biosecurity, Metrobiotech, Jumpstart Fertility, etc. In 2016 the first human clinical study for NMN was started in Tokyo by Hiroshi Itoh of Keio University, and in 2017 human trials of Sinclair's method began at a hospital in Boston. There are already commercially available NR supplements that in theory can boost your NAD, but these are the first clinical studies. David Sinclair has discovered that, by raising the levels of the molecule NAD in the blood of mice (via the administration of nicotinamide mononucleotide or NMN), the aging of blood vessels can rapidly be reversed. His study seemed to confirm that the decline of NAD over the years causes the loss of the protein Sirtuin ("Impairment of an Endothelial NAD+-H2S Signaling Network Is a Reversible Cause of Vascular Aging", 2018).

In 2015 Stuart Kim's team at Stanford University (notably Kristen Fortney, who immediately founded BioAge Labs in Berkeley) conducted a study of the genetic data of centenarians. They used data for the 801 centenarians listed in the New England Centenarian Study (NECS), the largest centenarian GWAS, and data for the 5,406 elderly over age 90 of another famous study. This led them to focus on eight gene variants that seemed to be shared by many of these long-living people. Then they examined the genetic data for 474 centenarians of Ashkenazi Jewish descent, 440 centenarians from Southern Italy and about 226 from Northern Italy (NICS). This narrowed down their list to four variants: APOE/TOMM40, which is associated with Alzheimer’s disease; CDKN2B, which is associated with cancer cell division; ABO, which is related to the type-O blood group; and SH2B3/ATXN2, that extends the lifespan of fruit flies.

Potentially, this is big business, so no surprise that Craig Venter opened Human Longevity Inc (in San Diego in 2013) and Jesse Karmazin started Ambrosia (in Monterey in 2016) to investigate the finding by Tony Wyss-Coray at Stanford University that a transfusion of younger blood makes old mice live longer. Ambrosia launched its first clinical trial in June 2016, which concluded in January 2018, and we still don't know the results. In 2014 Tony Wyss-Coray himself co-founded Alkahest in Redwood City.

In 2011 Jean-Marc Lemaitre's lab in France took cells from centenarians and reprogrammed them in the lab to "rejuvinate" them ("Rejuvenating Senescent and Centenarian Human Cells by Reprogramming through the Pluripotent State", 2011). That opened the door to another line of research: artificially inducing the "Yamanaka factors" (Oct4, Sox2, C-myc and Klf4) to reset the aging clock. Based on studies by Rando and Howard Chang at Stanford ("Aging, Rejuvenation, and Epigenetic Reprogramming", 2012), suggesting how the aging clock could be reversed, in 2016 Juan-Carlos Izpisua-Belmonte at the Salk Institute in San Diego succeeded in reprogramming cells in mice to express the four Yamanaka factors. These mice lived more than 30% longer.

Other startups in the field of "extended longevity" are resTORbio, which is a Novartis spinoff, Insilico Medicine, Mount Tam Biotechnologies, and Gero in Russia. Rejuvenate Bio, cofounded in 2016 by George Church, has a mission to make us " live to 130 in the body of a 22-year-old". It is experimenting to rejuvenate dogs using gene therapy. resTORbio, headed by Chen Schor, was established in Boston to commercialize the results of a Novartis study that ran from 2015 till 2017 in Australia and New Zealand: they tested a substance similar to rapamycin called everolimus on elderly people and the results are encouraging.

Startups in the field of regenerative medicine include Samumed, founded in 2008 in San Diego as Wintherix by Dennis Carson (who had already founded Triangle Pharmaceuticals and sold it to Gilead) and Osman Kibar, a startup that emerged in 2018 with a lot of funding and several drug therapies that could rejuvinate various parts of the body (the biggest investment that Singapore-based venture capitalist Finian Tan has made since he invested in a young Chinese startup called Baidu); and Celularity, founded in 2018 in New Jersey by Robert Hariri (also cofounder of Human Longevity), which is using placenta stem cells to regenerate tissues.

Genetics and big data are also enabling a new science of predicting people's life span. For example, in 2017 Yan Zhang at the German Cancer Research Center introduced a "mortality risk score". Jim Wilson's group at the University of Edinburgh introduced a similar scoring system for predicting mortality based on DNA. Steve Horvath at UC Los Angeles published a tool called DNAm GrimAge that calculates your life expectancy and a tool called AgeAccelGrim that calculates the year when you will develop cancer or have a heart attack. Of course, there are ways to beat the predictions, but the predictions is useful to make you realize that you have a serious problem, and you shouldn't wait until the last day to solve it.

Gene therapy is likely to be the most effective way to prolong a person's life. Exercising and diet influence how long you will live, but nothing compares to the influence of your genes. Unfortunately gene therapy is likely to remain extremely expensive, so these studies raise the issue of who will be able to afford these "longevity" drugs that cost half a million dollars: only the very rich. This is not a new fact. A 2012 study from Health Affairs showed that rich people are likely to live 12 years longer than poor people in the USA. The life expectancy of college-educated white men was about 80 whereas the life expectancy of white men without high-school education was 67. Among white women the difference was 84 to 73. The numbers are even more worrying if you compare them with previous generations: poor people actually die younger today than in the past. For example, the life expectancy of white women without high-school education was five years shorter than those of white women without a high-school degree of one generation earlier (18 years earlier). Gene therapy will simply increase the gap. Rich people may be able to live to the age of 120, whereas poor people may die before 60.

Anti-aging treatments are offered by hundreds of clinics but here are all unregulated therapies. Meanwhile the record of longevity still belongs to Jeanne Calment, who died in 1997 at the age of 122. Nobody has managed to live longer.

The fact that may disturb you is that these DNA testing companies make money out of your genes. Genomic companies are ammassing a treasure: the genomes of their customers. The customer of 23andMe or Ancestry is paying for the DNA test in two pays: with some money and with information about her genome. Genomic companies carefully collect the genomes of all their customers: that will be valuable information to analyze who has the "best" genes for some application. Everybody's favorite application is longevity. Companies that are looking for the longevity genes need to collect genetic samples from millions of people and then check which genes are prevalent in the people who live to be 100. The media paid a lot of attention when in 2012 Amgen purchased the genomes of 160,000 inhabitants of Iceland, but that's nothing compared with the database of genomes that have been amassed by Ancestry and by 23andMe, which has more than one million genomes. And of course when Google entered the fray with Calico people immediately thought of Google's ability to collect "big data".

There is already a gold rush for "rare genes", for genetic mutations that give some individual some advantages that others may want. For example, some individuals were publicized by the news media in the USA because they don't feel pain. That's actually a disease, because their lives are constantly in danger: pain is a useful signal from our body that something is wrong, and you can die if you don't get that signal; but, if we could understand which genetic mutation causes that "disease", we would be able to create a new family of pain-killers. Those "rare genes" are worth millions. They belong to people who took the DNA test and who will not receive a penny from the inventions due to their rare genes.

On the other hand, the information about our genomes can be valuable to save lives. Rare genes are interesting, but also important is to study "rare diseases". Rare diseases are not well understood because few people have them. If these people volunteer their genomes, scientists can work on a much bigger dataset.

Let's face it: if you get your genome sequenced, you can do very little with it. You do it mainly for fun. It is "entertainment", not "health care". The reason that it is not very useful is that scientists know too little about the correlation between genes and diseases. We need to have millions of genomes and for each one we need data about the person's health. Only then will we be able to find meaningful and useful correlations. This has become a "chicken and egg" problem. My motivation to get my genome sequenced is very low because i know that it is not very useful; but it is not very useful because we have very few genomes sequenced and data about these people's health. At this point i think that protecting my privacy is not a high priority. The high priority is to convince me to get my genome sequenced and then provide information about my health for the rest of my life, so that i can help scientists find correlations and make the test more useful in the future. At this point society should invest more in motivating people to get their genome sequenced than in protecting their privacy.

In 2012 Britain launched the 100,000 Genomes Project via a company called Genomics England: people with rare diseases can upload their genomes via an app, PanelApp, to help scientists who study the genetics of rare diseases.

Scientists are increasingly interested in studying populations of genomes, not just individual genomes. The Human Genome Project was a big success and delivered a "blueprint" of how the human software works; but we are all different: there are genetic variations between person and person. Those genetic variations can make the difference between living a healthy life and dying of cancer at a young age.

The Personal Genome Project is an interesting merger of crowd-sourcing and biotech ideas. Originally launched in 2005 by George Church at Harvard University, its goal was to enroll thousands of volunteers willing to have their complete genomes and medical records published on the Internet, so that researchers all over the world could study them and find correlations among genes, environment and diseases. By 2015 the project had enrolled more than 16,000 volunteers.

In 2008 David Altshuler of the Broad Institute of Boston, which is a joint laboratory of the MIT and Harvard University, and the National Human Genome Research Institute (NHGRI) in Maryland launched the 1000Genomes project (www.1000genomes.org) to study human genetic variation. Several laboratories helped, including BGI-Shenzhen. This project collected the genomes volunteered by 1,000 people from all over the world and analyzed the differences. But 1,000 is really a drop in the ocean when we know that millions of people have already sequenced their genomes. In 2014 one of the project's scientists, the Israeli-born computational biologist Yaniv Erlich, moved from the MIT to the New York Genome Center, a spin-off of Columbia University, where in 2015 he and Joe Pickrell launched DNA.land (http://dna.land), another non-profit project to collect people's genomes and study genetic variation. This time it was truly a case of "crowdsourcing": asking people from all over the world to upload their DNA so that scientists can study it. In 2015 Yaniv Erlich published the paper "A Vision for Ubiquitous Sequencing" in the Genome Research magazine.

The crowdsourcing experiment then migrated to the West Coast, where people started talking about the "Internet of DNA" or "Internet of Living Beings" (in contrast with the "Internet of Things"). The hero in California is David Haussler of the University of California in Santa Cruz, who in 2013 partnered with David Altshuler of the Broad Institute to create the Global Alliance for Genomics and Health. The idea is to set up a peer-to-peer network of scientists and volunteers to work together on understanding genetic variations. In 2015 the MIT Technology Review had an article about this project: "A global network of millions of genomes could be medicine's next great advance."

In 2015 WuXi AppTec (founded in 2013 in Wuxi) acquired one of the largest genomic databases for rare diseases and cancers, NextCode, built over twenty years by Decode Genetics, founded in 1996 in Iceland to manage and mine the first truly massive collection of genomic data. In 2017 iCarbonX (started in 2015 in Shenzhen by Jun Wang, a former BGI executive, and mostly funded by Tencent) invested $100 million in PatientsLikeMe (founded in 2004 in Boston) that holds the world's largest database of patients with rare and chronic diseases, DigitalMe (and this was the seventh such "alliance" formed in just two years with an international firms). PatientsLikeMe is a patient network that enables patients to connect with people who have the same condition. In the process they generate valuable data about their condition. (The goal of iCarbonX is to achieve full-body digital simulations). NuMedii was founded in 2008 in Silicon Valley specifically to commercialize a huge database of annotated data owned by Stanford University about human, biological, pharmacological and clinical events.

And finally in 2016 a big pharmaceutical company, AstraZeneca, launched a project to sequence 2 million genomes (in collaboration with Venter's new startup Human Longevity, Britain's Wellcome Trust Sanger Institute, and Finland's Institute for Molecular Medicine).

A simple example of the benefits of these databases came in 2016. From 2006 until 2010 British scientists have collected blood, urine and saliva samples from 500,000 adult volunteers for the project UK Biobank, hosted at the University of Manchester and directed by Rory Collins of the University of Oxford. The scientists keep monitoring the health of these volunteers. Using those data, in 2016 the University of Edinburgh identified two genetic variants that can shorten a person's life by 3 years. This finding affects about 3 in 1000 people.

In 2011 the National Research Council of the USA published a report titled "Toward Precision Medicine - Building a Knowledge Network for Biomedical Research and a New Taxonomy of Disease." In 2015 the US government launched the "Precision Medicine Initiative", whose goal is to catalog the genomes of one million people. Precision medicine is about customizing treatment based on individual genes. The goal is "pharmacogenomics": the right drug for the right patient at the right time and at the right dose. The idea behind it is that some genes cause a predisposition to some diseases. The only way to find out if this theory is correct is to find the genes shared by people with certain diseases. It has been more than ten years since the human genome was sequenced, but we don't really have a success story in precision medicine. In 2012 the USA approved a drug by Vertex, Ivacaftor, for curing cystic fibrosis, one of the most expensive drugs ever developed by the pharmaceutical industry, but the results have been disappointing: other (more traditional) treatments that cost a fraction of ivacaftor seem to achieve the same results. Precision medicine needs big data that today are still missing. The 2011 report encouraged two new repositories: an "Information Commons" that will make data on large populations of patients available to all scientists; a "Knowledge Network" that will highlight inter-relationships. An important tool is the GWAS (Genome-wide Association Study), that tries to relate common genetic variants in different individuals to traits. The GWA catalog (originally maintained by the National Human Genome Research Institute or NHGRI, but now hosted by the European Molecular Biology Laboratory-European Bioinformatics Institute or EMBL-EBI) was set up in 2008. It is a database that associates "single nucleotide polymorphisms" or SNPs to traits.

Precision medicine could also help "predictive" medicine. Today most of the money spent by the health-care system is spent after people get sick; and most of the cost to the government and to society happens after people get sick. It would be nice to reverse this statistic, and get to the point when most of the money is spent in preventing the disease. That will be possible when we know which genetic variations are more likely to cause a disease.

We also need a more fair sample of genomes. The vast majority of people who have had their genomes sequenced so far are of European ancestry. Therefore, the millions of genomes available to the genomic startups are not very useful to understand non-white people.

In 2016 Mark Zuckerberg donated $600 million to BioHub, that hired Stephen Quake of Stanford University as president. BioHub's goal is to create a map of the millions of cells in the human body.

Affymetrix introduced the first "DNA chip" in 1994, the GeneChip, based on photolithography, thanks to the 1991 breakthrough by its founder Stephen Fodor when the company was still called Affymax. Affymetrix introduced the first DNA chip in 1994, the GeneChip. Pat Brown and Mark Schena at Stanford University worked on a different method (a robotic method) and in 1995 introduced the term "DNA microarray". Edwin Southern at Oxford University (and the founder of Oxford Gene Technology in 1995) was working on a technique based on inkjet printing, and so did Alan Blanchard at the University of Washington, who in 1996 invented the technique adopted by Agilent. Nimblegen Systems adopted an improved version of Affymetrix's technique. Illumina adopted the method invented by David Walt at Tufts University in 1998. They all wanted to leverage techniques originally developed for silicon semiconductors in order to improve the speed at which DNA tests could be performed. Their microarrays made it possible to simultaneously test thousands of molecules. The microarrayers that performed such DNA tests were descendants of the first array robots built in 1987 by Hans Lehrach at the the Imperial Cancer Research Fund in England.

The next step in biotech automation after the DNA chip/microarray was the "lab on the chip". Since the 1960s there had been a lot of progress in "micro-electro-mechanical systems" (MEMS). These devices were already around before the invention of the microprocessor. In 1964 Harvey Nathanson at Westinghouse made the first MEMS, and the first success story of MEMS was the "thermal inkjet" technology that Hewlett Packard debuted in 1979, followed in 1993 by Analog Devices' micro-accelerometer (widely used in many industries today, for example in airbags). In 1983 Richard Feynman delivered one of his famous lectures, this one titled "Infinitesimal Machinery". Initially MEMS simply exploited the fabrication technologies of the semiconductors industry, but in 1999 Lucent introduced the all-optical router and triggered the boom of optical MEMs of the 2000s. But the enabling technology was "microfluidics", the ability to make millions of microchannels ("micro" in the sense that they measuring micrometers in diameter) that handle very tiny quantities of fluids. This was the result of a US military program: the Defense Advanced Research Projects Agency (DARPA) wanted a system to quickly detect biological and chemical weapons and so in 1997 created a program called "Microflumes" to fund research in microfluidics. In 1978 James Angell at Stanford had already been working on "micromachines" and one of his students, Stephen Terry, in 1979 had unveiled what can be considered as the first "lab on a chip", a device for separating, identifying and analyzing the components of a gas (originally, it was commissioned by NASA and meant to analyze the atmosphere of Mars. But progress in MEMS and in microfluidics led to today's "lab-on-a-chip" products. In 1999 Hewlett-Packard's spinoff Agilent introduced the first commercial "lab-on-a-chip" product, the 2100 Bioanalyzer. Even more important was the Agilent 5100 of 2004. These were the vanguard of systems that enabled biotech startups to conduct analysis of thousands of DNA and protein samples per day. After the success of the Human Genome Project, the goal shifted to putting the whole human genome on a microarray. In 2002 Wilhelm Ansorge at the European Molecular Biology Laboratory (EMBL) in Germany succeeded.

In 2004 the first commercial microarrays with the whole human genome became available from Affymetrix (whose GeneChip was still dominating the market for microarrays), Agilent (that still relied on the technique based on inkjet printing), Applied Biosystems and Illumina (all based in California, the first three in the Bay Area). Technically speaking, the first company to offer a whole-human-genome microarray was probably Wisconsin-based NimbleGen Systems in 2003. Then the battle began for lower prices and for better "annotation" of the genes. In 2009 Arrayit, founded in 1993 as TeleChem International in Sunnyvale by Rene Schena and Todd Martinsky, introduced the H25K.

There is probably going to be more progress because interest in "labs on a chip" is enormous both in the industry and in the government. For example, in 2011 DARPA launched another program, this time to create "human-on-chip" platforms, i.e. more comprehensive and evolved "labs on a chip".

For most biotech tasks we still need to handle liquids in a real-world laboratory. This can be done by an expensive human, who works only some hours and on some days, or by a robot, who works nonstop. There are already robots that can do what a biologist does manually, but they cost $100,000 and more. The goal now is to lower that cost so that small laboratories can afford it. OpenTrons is an interesting story. It was founded in 2013 by Will Canine, a graduate from New York University who was also an activist of the anti-capitalist "Occupy" movement and a "biohacker" at the Do-It-Yourself biotech space Genspace, with the Chinese robotics expert Chiu Chau and the software engineer Nick Wagner. OpenTrons was incubated in Shenzhen in 2014 by Haxclr8tr and initially funded via a Kickstarter campaign. It is now based in New York, Their goal is to build an affordable robot for performing biotech experiments. Their project is clearly inspired by the software hackers of Silicon Valley: it is open-source and provides a form of "rapid prototyping", except that they are not prototyping software but life forms. OpenTrons' robot is built around an open-source Raspberry Pi computer (originally conceived in Britain in 2012 for schoolchildren by a charitable organization based around Cambridge University) and open-source software. They hope to make it cheaper than a laptop computer so that the DIY community can use it.

The other promising technology is the cloud. Cloud-based biotech completely eliminates the laboratory for the customer. Transcriptic, founded by Duke University graduate Max Hodak in 2012 in Palo Alto, conducts laboratory tests using robots on behalf of scientists who can be located anywhere in the world. They have the robots, they have the laboratory, they handle all the computation. You simply submit the specifications of your experiment and they perform it on your behalf.

In 2014 Emerald Therapeutics (2010, San Francisco) introduced the Emerald Cloud Laboratory, a laboratory on the cloud where robots conduct the experiments. The scientists use a symbolic language to "program" the experiments.

The software side of recombinant DNA technology is becoming more important now that the hardware is no longer the main obstacle. The manufacturing sector uses rapid prototyping and CAD/AM software to design products. Biotech needs something similar to "design" genetic experiments. The US government set up three main research centers for biotech: the BioEnergy Science Center, led by Oak Ridge National Laboratory; the Great Lakes Bioenergy Research Center, led by the University of Wisconsin-Madison and Michigan State University; and the Joint BioEnergy Institute (JBEI), led by Lawrence Berkeley National Laboratory (Berkeley Lab) in the Bay Area. Nathan Hillson, the director of the synthetic biology program at JBEI, built the j5 software package for biotech design and then in 2011 founded a startup, TeselaGen Biotechnology, to commercialize it.

We are moving towards completely digitazing the laboratory. Synthace (2011, Britain) has developed Antha, an operating system and programming language for laboratory scientists, with the aim to digitize the entire process of drug discovery and manufacturing.

We can envision a day when individuals will be able to program a living organism on a smartphone, upload the design to the cloud and order the living organism from a laboratory that will download the design from the cloud and use robots to manufacture the living organism.

If you think like a computer scientist, by 2010 biotech had reached a point where it was easy to read the genetic data (DNA sequencing) and easy to write new genetic data (DNA synthesis) but it was still difficult to edit genetic dada (the genome). One of the earliest methods, the ZFN method, was exclusively owned by Sangamo Biosciences.

The history of genome editing begins with Oliver Smithies at the University of Wisconsin-Madison and with Mario Capecchi at the University of Utah, who pioneered a technique called "homologous recombination" (respectively in 1985 and 1986). Initially, their method was only feasible in laboratory cells, not in living beings, but in 1987, both Capecchi and Smithies achieved the modification of a gene in a live mouse. Meanwhile, in 1985 Aaron Klug's team at Cambridge University identified protein structures called "zinc fingers" in frog eggs, and today we know that they are common in all living beings (except bacteria). Zinc fingers are useful for a variety of tasks inside the cell. In 1994 Maria Jasin's team at the Sloan Kettering Institute in New York figured out how to make "double-strand breaks", and in 2001 Dana Carroll at the University of Utah figured out how to make such "breaks" anywhere in the genome using zinc-finger nucleases, the artificial enzymes whose power for editing DNA had been discovered in 1996 by Srinivasan Chandrasegaran at Johns Hopkins University (Johns Hopkins University licensed the technology exclusively to Sangamo Biosciences). ZFNs were the vanguard of programmable nucleases (later additions are TALENs and CRISPR-Cas9), enzymes that are useful for delivering a double-strand break at a precise location. After Carroll's demonstration of the ZFN method in frog cells (2001) and in cells of the fruit fly (2002), David Baltimore and his student Matthew Porteus at CalTech demonstrated it in human cells (in 2003), followed by Fyodor Urnov's team at Sangamo Biosciences (in 2005).

The capability of TALE proteins to bind to DNA was first described in 2007 by Sebastian Schornack and others in Thomas Lahaye's team at Martin-Luther University in Germany. The TALEN method invented in 2008 by Dan Voytas and Feng Zhang at the University of Minnesota (and widely available after Adam Bogdanove of Iowa State Univ and Voytas published the TALEN kit in 2011) was much faster, and hailed as dramatic progress. The first demonstration in cells came in 2010, with the construction of TALEN nucleases in yeast by Daniel Voytas' team at University of Minnesota in collaboration with Adam Bogdanove's team.

But in just one year an even better, easier, cheaper and faster technology emerged: Jennifer Doudna's laboratory at UC Berkeley and Emmanuelle Charpentier's laboratory in Sweden invented the CRISPR-based technique for gene editing. Note that the discovery came by accident. Doudna and Charpentier were studying how bacteria protect themselves from viral infections: CRISPR is their sophisticated RNA-guided defense system.

Feng Zhang's team at the Broad Institute was probably the first one (in 2011) to make it practical to edit human cells (using TALEN). CRISPR stands for "clustered regularly interspaced short palindromic repeats". It consists of chunks of DNA that repeat a certain sequence.

Many people contributed to the "invention" of CRISPR editing. And it is a classic case of scientific serendipity: it all started with a fight between corporations over how to make the best yogurt. By sheer accident, scientists discovered that nature had already invented a gene-editing tool: they found it in yogurt. It was first described in 1993 by Francisco Mojica (in Spain) and in 2005 the same Mojica realized that CRISPR is an adaptive immune system. In 2005 Alexander Bolotin (in France) discovered Cas9. In 2006 Eugene Koonin (at the NIH) showed that the Cas proteins can be used to repair DNA. In 2007 Philippe Horvath's team at Danisco (France) proved that CRISPR systems are adaptive immune systems, but Danisco only used the discovery to make yogurt. Because the giant DuPont acquired Danisco in 2011, many people eat yogurt enhanced with CRISPR every day. In 2008 John van der Oost (Netherlands) guessed how to program CRISPR to target DNA and at the end of 2008 by Luciano Marraffini and Erik Sontheimer (University of Illinois) showed that CRISPR can serve as a general-purpose genome-editing. In 2010 Sylvain Moineau in Canada, Emmanuelle Charpentier, who was then in Sweden, and Joerg Vogel in Germany focused on the Cas9 nuclease. Moineau showed that CRISPR-Cas9 creates double-stranded breaks in target DNA at precise positions. In 2011 Virginijus Siksnys in Lithuania figured out how to design a CRISPR-Cas9 system to cut DNA in test tube, followed in 2012 by Charpentier and Jennifer Doudna of UC Berkeley (but the paper of the latter came out before Siksnys' paper). Doudna and Charpentier (Sweden) used a CRISPR-Cas9 system to cut DNA in test tube. A few months later (the paper was published in January 2013), Feng Zhang and Luciano Marraffini at the Broad Institute invented the first CRISPR system to edit human cells (eukaryotic cells in general). In April 2013 Chad Cowan and Kiran Musunuru of Harvard University published a study showing that CRISPR was better than the other gene-editing tools. The story of the CRISPR technique is interesting because most of the protagonists were born outside the USA and away from the famous universities, several of their papers were rejected by journals, and most of them were in their 20s.

CRISPR startups, offering "genome-editing platforms", started popping up everywhere. The first one, in 2011, was Rachel Haurwitz's and Martin Jinek's Caribou Biosciences, a spinoff of Doudna's lab at UC Berkeley, but within a few years similar startups spread from Switzerland (CRISPR Therapeutics, founded in 2013) to Boston (Editas Medicine, a 2013 spin-off of the Broad Institute, and Intellia Therapeutics, founded in 2014 by Caribou itself). In 2015 scientific journals published more than 1,300 papers on CRISPR.

These startups have been formed by the CRISPR protagonists in countless combinations: Charpentier and Cowan founded CRISPR Therapeutics; Zhang, Church, and Doudna founded Editas; Sontheimer, Marraffini, Derrick Rossi, Barrangou and Doudna founded Intellia Therapeutics; etc.

By 2019 three CRISPR startups had gone public (CRISPR Therapeutics, Editas and Intellia). In November 2019 Emmanuelle Charpentier`s CRISPR Therapeutics announced that the first patient to receive their experimental CRISPR-based treatment CTX001 for sickle-cell disease is doing well: Victoria Gray could become the first person ever healed of the disease.

CRISPR editing has been used to create new fruit (Joyce Van Eck at Cornell Univ in collaboration with Zachary Lippman at the Cold Spring Harbor Laboratory), to change the color of flowers (at the University of Tsukuba in Japan), to heal muscular dystrophy (at the University of Texas Southwestern), to treat sickle cell disease (Emmanuelle Charpentier's CRISPR Therapeutics in collaboration with Joshua Boger's Vertex Pharmaceuticals), to eliminate mosquitoes (Andrea Crisanti at Imperial College London), and so on.

Cas9 is an "RNA-guided DNA cutter". After Cas9, other similar "cutters" have been identified: Cas12a, Cas12b, etc And there are also several RNA-guided RNA cutters: Cas13a, Cas13b, etc.

The CRISPR-Cas9 system has also been used to the CRISPRi (CRISPR interference) and CRISPRa (CRISPR activation) methods. CRISPRi silences genes (similar to RNAi, a natural process discovered in worms in 1998), and CRISPRa activates genes. The CRISPRi technique, that uses a "dead" Cas9 (dCas9) protein, was developed by Wendell Lim's student Lei "Stanley" Qi at UC San Francisco ("Repurposing CRISPR as an RNA-guided platform for sequence-specific control of gene expression", 2013).

Genome editing is a powerful and scary technology. It can be "somatic" cell editing, which is non-heritable, or "germ" cell editing, which is heritable. The former is scary enough, but the latter is even passed down to all future generations. Just two years after the invention of the CRISPR method, in 2014, Weizhi Ji's team at Kunming University in China used CRISPR to edit the germ-line of monkeys (i.e. the changes will be passed on to future generations).

Using CRISPR-Cas9 we already had some major innovations in gene editing. In 2016 David Liu's team at Harvard University improved over CRISPR with a new technique called "base editing". You can view the genome as consisting of 20,000 genes or as consisting of six billion DNA letters. Each gene is made of DNA and DNA is expressed in an alphabet of four letters, or chemical bases: adenine (A), cytosine (C), guanine (G) and thymine (T). The DNA of a person is written in two strings whose letters have to match. Some genetic diseases are caused by an error in the pairing of letters within the DNA of the person (by "point mutations"). Liu discovered a method to edit the single base/letter, i.e. to correct single-letter errors in a person's DNA ("Programmable base editing of AT to GC in genomic DNA without DNA cleavage", 2016). CRISPR targets an entire gene. It relies on "double-stranded DNA breaks" that cause a stochastic disruption of the entire gene: "stochastic" means "random". Several genetic diseases (notably, sickle-cell anemia) are not caused by an entire gene: they are caused by a simple mutation inside a (useful) gene, a "point mutation". In those cases, CRISPR is overshooting. Let's say that you have a fleet of trucks and one has a flat tire: you don't throw away the whole truck, you change the tire. Liu's lab created two different kinds of base editor: cytosine base editors, which convert a C nucleotide to a T, and adenine base editors, which convert a A nucleotide to a G. Liu co-founded two startups to commercialize his technology: Pairwise Plants and Beam Therapeutics. In 2019 Hui Yang's team at Shanghai Institutes for Biological Sciences found a problem with Liu's cytosine base editors (but not with adenine base editors) that slowed down implementation of this technique. In 2017 Feng Zhang's team at the MIT invented a method called REPAIR (RNA Editing for Programmable A to I Replacement") to edit not DNA but RNA, which turns DNA's code into the genetic instructions to make proteins ("RNA editing with CRISPR-Cas13", 2017).

CRISPR's popularity went both ways. On one hand, the technology was rapidly improving. Scientists had been struggling for decades to reprogram T-cells. The traditional technique was to use "viral vectors" to deliver DNA to the cells, but this was complicated. In 2018 Theo Roth in Alex Marson's laboratory at UC San Francisco discovered a fast, simple CRISPR-based method for reprogramming T-cells that used a different technique, "electroporation" instead of viral vectors ("Reprogramming Human T-cell Function and Specificity with Non-viral Genome Targeting", 2018). In 2020 Carl June's team at the University of Pennsylvania disclosed research on reprogramming the receptor of T cells (located on the surface of the cell) to improve the cell's ability to detect and kill tumor cells ("CRISPR-engineered T cells in patients with refractory cancer", 2020). On the other hand, there were doubts about the risks of CRISPR. First,n 2017 a joint team of Stanford University, the University of Iowa and Columbia University published a study ("Unexpected Mutations After CRISPR-Cas9 Editing In Vivo", 2017) that CRISPR may create thousands of unintended mutations. within a few months their study was proven faulty and they published a retraction, but the damage was done: they had created alarm in the public. Then in 2018 a study by Allan Bradley's team at the Wellcome Sanger Center in Britain described how CRISPR can lead to massive genetic mistakes: chunks of DNA being deleted, reversed, or misplaced ("CRISPR Gene Editing Produces Unwanted DNA Deletions", 2018).

CRISPR looked great on paper, but its "molecular scissors" were difficult to use for precise editing in most cell types. In 2019 David Liu's student Andrew Anzalone at the Broad Institute developed a technique for editing genes using the same Cas9 nuclease but without using double-stranded DNA breaks (like CRISPR does): prime editing ("Search-and-replace Genome Editing Without Double-strand Breaks or Donor DNA," 2019).

DNA synthesis itself is being revolutionized by the combined forces of miniaturization, automation and software. All the companies that want to do "rapid prototyping" in polymerase chain reaction (PCR) or gene sequencing need some raw materials called oligonucleotides. These are short DNA molecules that machines can use to test hypotheses. A traditional limit was the cost of these "oligos". A hot startup in San Francisco is Twist Bioscience, founded in 2013 by Emily Leproust, a veteran of Agilent, by Complete Genomics's hardware engineer Bill Banyai and by Bill Peck, who worked at both Complete Genomics and Agilent. They want to produce synthetic DNA on a massive scale using a silicon-based method to make oligos. In 2015 Twist Bioscience got an order of 100 million base pairs of DNA (the equivalent of millions of oligos) from Ginkgo Bioworks that was about 10% of the world's global gene market. In 2016 they acquired Israel's Genome Compiler that has a technology to design genes, so now Twist can let people design genes and then it can print them on demand. Meanwhile, the largest gene-synthesis supplier in the world is Nanjing-based GenScript that provides custom gene-synthesis for scientists thanks to a sort of assembly line for oligo synthesis. Right now synthetic biology is still using "cut and paste" technology to create new organisms; but, as costs continue to decline, some day it might be more effective to print a new genome rather than edit an existing one.

In 2014 a team led by Jef Boeke at Johns Hopkins University for the first time succeeded in synthesizing (creating) a chromosome in a complex organism (only yeast, but still a major step towards creating living organisms). The final goal is to create the world's first artificial genome, the genome of the yeast (16 chromosomes for a total of about 5,000 genes). The project is now distributed around the world. For example, Junbiao Dai, Boeke's collaborator at Tsinghua University, is synthesizing the 12th chromosome, the largest of the 16th and the one They have developed new techniques that could be more important than their official goal. For example, they invented "gene scrambling", which is a mix of synthesis and editing: it creates a new organism out of the same genes. This can be useful to simulate what happens when some genes are not in the right place. Boeke's collaborator Patrick Cai at the University of Edinburgh is trying to add a 17th chromosome to the 16, a chromosome that should add a new function to the yeast. Boeke's student Jingchuan Luo is trying to pack all 16 chromosomes in just one. It is not clear if these things are possible: is there a minimum number of chromosomes for the yeast? is it possible to add a function to an existing organism? The genome of the jumper ant of Australia (myrmecia pilosula) is made of just one chromosome. The blue butterfly (polyommatus atlanticus) is the species with the highest chromosome number: at least 223. Why some animal species have only one chromosome and some have hundreds?

GMO also suffers from another form of bad publicity: it has helped the big chemical corporations make billions of dollars.

The obvious advantage of creating new food in the laboratory is that it takes only a few months instead of 10-20 years. By definition, the plant created inside the laboratory is also more "scientific" than the plant created by the farmer by "trial and error": the scientist knows why the plant grows, whereas the farmer only knows that his experiment either it worked or didn't work, but doesn't really understand why.

This is also what scares people. People assume that the traditional process, precisely because it takes so long to create a new variety of potatoes or tomatoes, is less dangerous: there is enough time to test it, a little bit at a time. This may be true, but there is a simple reason why we cannot afford to wait 10-20 years to improve our plants. Climate change is happening faster and faster. We need to adapt our crops and fruit to the rapid fluctuations of climate change, and genetic engineering can do that. Climate change is bringing higher temperatures to places where plants need a cold season, and too much rain to places where plants cannot defend themselves against the parasites and diseases that spread with humidity. There seems to be evidence that catastrophic storms are becoming more frequent. Climate scientists tell us that extreme weather will become a common occurrence, not a rarity. For a farmer this means that the weather is becoming too unpredictable. David Lobell at Stanford and Wolfram Schlenker at Columbia University have published studies about the effects of climate change on food production, such as "Climate Trends and Global Crop Production Since 1980" (2011) that showed a decline to corn and wheat production due to climate change. Think of rice, that is the main ingredient of the diet of about 40% of the world's population. If rice succumbs to climate change, we will have starvation again in countries like India and China. In 2009 British scientist Paul Quick from the University of Sheffield was appointed to run the "C4 Rice Project" at the International Rice Research Institute (IRRI), a consortium of 12 laboratories in 8 countries. This project, funded by the Bill & Melinda Gates Foundation. This project aims to improve rice with a technique called "C4 photosynthesis". Meanwhile, Eduardo Blumwald, a scientist from UC Davis, is experimenting in California's Central Valley, one of the world's most productive agricultural regions that has recently experienced extreme drought and heat conditions. Blumwald wants to engineer plants like rice that can tolerate extreme variations in heat and drought, and that can grow in soils with high salinity.

In California there is also a "green" movement to redesign the food supply so that it uses less land, water and energy. It turns out that meat is the worst food that humans can possibly eat in terms of land use, water use and energy use. So there is also research into creating plant-based substitutes for animal-agriculture products, in other words "fake meat". A startup that became very popular in San Francisco is Hampton Creek, founded in 2011 by social worker Josh Tetrick and animal-rights activist Josh Balk. Their animal-free replacement for the chicken egg is sold in many stores and their egg-free "Just Mayo" mayonnaise is used in many sandwich places. They were viciously attacked by some big lobbies and corporations, but they won their legal battles. Egga are a big market: $60 billion in the USA alone. Venture capitalists are now funding several food-related startups. Hampton Creek is not a biotech company because it simply looks for a vegetarian way to provide the same nutritional value and taste of animal products, but its success story has inspired biotech companies to achieve the same results in the laboratory; and it has convinced venture capitalists that people will switch to plant-based substitutes if that option is available. At the end of 2020, Singapore became the first country to approve meat grown in a laboratory: it was chicken meat, created directly from animal cells by Hampton Creek, now renamed Eat Just.

The fear of genetically-engineered food remains, but it is largely based on superstition. Very few things have been subjected to so many studies like genetically-modified food. And yet, in more than 15 years, no transgenic crop has been found dangerous for humans. The majority of corn, soybeans and cotton produced in the USA have been created in laboratories using a gene from a bacterium. 81% of the soybeans of the world are genetically modified. 96% of India's cotton is genetically modified. Even better: the new TALEN and CRISPR techniques can modify a plant without the addition of genes that come from other living beings. They offer an easy and precise way to edit (alter) the plant's genes so that it will be drought-resistant or something else. In theory, this should reduce the chances of "transgenic" catastrophes. If you object to editing the genes of a plant, then you should also object to gene therapy for humans. They are, ultimately, the same process.

In fact, the US Department of Agriculture is still undecided whether crops created with the CRISPR and TALEN technique should be considered GMOs (Genetically Modified Organisms). I am sure that many groups will pressure the USA to classify these foods as GMO, but the fact that it is not so clear tells you that even the traditional reasons to fear GMOs are not there anymore.

Using genome-editing tools like TALEN and CRISPR, scientists can "genetically engineer" a lot more vegetables. For example, in 2014 Gao Caixia in China used both TALEN and CRISPR to create a strain of wheat that is fully resistant to powdery mildew. This experiment was followed by genetic engineering of tomatoes, soybeans, rice and potatoes.

But for farmers a "gene drive" technique might be even more important than a "gene-modification" technique. Genetic changes usually take many years of even millennia to spread in a population. A "gene drive" is a mechanism to spread a genetic change quickly inside a population. A "gene drive" could be used by farmers to wipe out undesired weeds. For example, we can (theoretically) genetically-change mosquitoes so that they will not spread malaria or genetically-change ticks so that they will not spread lyme disease, but then we have no way to make sure that this genetic change (these "good genes") will spread to all mosquitoes and all ticks of those dangerous species. Mendel's classical rules say that this is impossible in nature, but genome-editing changes the rules. The first experiments that seem to have succeeded in creating a "gene drive" was carried out in southern California. At the end of 2014 Ethan Bier and Valentino Gantz at UC San Diego used the CRISPR technique to trigger a gene drive in fruit flies. That was just a conceptual experiment, but a few months later Anthony James at UC Irvine in southern California used their experiment to introduce a "malaria-blocking" gene in a mosquito so that this new gene will spread to almost all its offspring. This scientist has spent 20 years of his life trying to create anti-disease mosquitoes in the laboratory. In 2018 Andrea Crisanti's team at Imperial College London carried out a similar experiment of a "gene drive" to exterminate mosquitoes using CRISPR to introduce a bad gene that makes it impossible for mosquitoes to reproduce.

Of course, this sounds even more terrifying to those who are afraid of genetic engineering. What is more terrifying to the mothers who live in places infested with malaria is that their children might die of malaria.

Before we criminalize genetically-modifed foods, let's remember that our foods today are not natural at all. Only chemical engineers can understand the labels of the foods at the supermarket. Read the labels, and you will find mysterious ingredients such as: butylated hydroxyanisole, butylated hydroxytoluene, polysorbate, sodium benzoate, sulfites, potassium sorbate, nitrates, etc. Our supermarket foods are full of artificial colors that make foods look pretty, artificial flavors that mimic the taste of natural ingredients, preservatives that prolong the shelf life of foods, and replacements for sugar like high-fructose corn syrup and calorie-free sweeteners. Some of these are possible carcinogens, some may weaken the immune system, some may cause allergies and possibly infertility, and some may even cause DNA damage. You would never eat these things if i asked you to eat them, but millions of people eat them every day. Since 1983 most cheese in the USA and Britain is made using chymosin (rennet) that has been genetically engineered, a much more humane method than the traditional method of obtaining it from the stomach of calves (a method still practiced in continental Europe). And, if you eat meat, you mostly eat animals that have been grown in industrial facilities and fed industrial food. Being afraid of genetically-modified foods when our diets are basically chemical experiments is a bit funny.

Narnia: What is the status of cloning animals?

piero:

These genome-editing technologies can have an even bigger impact beyond food. Perhaps the first successful application of the CRISPR technique was the experiment in 2014 by Chad Cowan and Derrick Rossi at the Harvard Stem Cell Institute. They took some human cells (some stem-cells that form blood and some "T-cells" that contribute to the immune system), edited them with the CRISPR technique, and then introduced the gene-edited cells into HIV patients. These "genetically engineered" cells are designed to fight the HIV. There is something in human cells that is vulnerable to HIV: remove that "something" and you get an HIV-resistant immune system. At the same time Daniel Anderson's team at the MIT used CRISPR (in mice) to correct a genetic mutation that causes a serious disease called "tyrosinaemia5" in mice. In 2015 JuanCarlos Izpisua-Belmonte's team at the Salk Institute in San Diego used CRISPR to remove HIV from infected cells before it could start replicating. Gene editing on a human being with CRISPR-Cas9 was first performed by Lu You at Sichuan University in Chengdu. These are the first steps towards using CRIPSR technology for gene therapy. Of course it will take years before these techniques are approved for use on human beings. Nobody today can predict the side effects.

There are a number of startups working on genetically-modified bacteria that humans could just swallow. In 2000 MIT's scientist James Collins turned a bacterium into the equivalent of an electronic "flip-flop" switch. Basically, he reprogrammed a bacterium the way a software engineer programs a computer. The startup that Collins founded in Boston, Synlogic, is engineering a bacterium capable of destroying ammonia. Ernest Pharmaceuticals (Boston) and GenCirq (San Diego) are engineering bacteria to treat cancer. Trayer Biotherapeutics (Maryland) is redesigning yogurt bacteria to treat a rare disease. ActoGenix (Belgium), acquired in 2015 by Intrexon, is engineering bacteria to treat allergic and autoimmune diseases.

We can even use these biotechnologies to create new materials. Without using these new technologies (that are still out of the reach of the average laboratory) several biotech startups have already created materials that didn't exist before. Refactored Materials (later renamed Bolt Threads), a 2009 spinoff of UC San Francisco (scientists Dan Widmaier, David Breslauer and Ethan Mirsky) manipulates bacteria to manufacture spider silk, stronger than steel but very light, that can be used for making clothes. Zymergen, founded in 2013 in Emeryville by two Amyris alumni, Jed Dean and Zach Serber, had discovered a way to insert DNA into bacteria, and to create microbes that can create new materials. It then uses robots that explore and tinker with each microbe’s genome until the desired material is produced. Modern Meadow, started in New York in 2011 by Gabor Forgacs of the Univ of Missouri, wants to "print" meat and leather without killing animals.

A new field is born: sustainable fashion. There are bacteria-produced dyes (Ginkgo Bioworks), Synthetic leather (Modern), kelp-based textiles (AgilKnit), and synthetic spider silk (Bolt Threads). Ginkgo Bioworks has even a designer-in-residence, Natsai Chieza.

The startup that makes synthetic biologists dream is Ginkgo Bioworks, founded in 2008 in Boston by MIT's synthetic biology pioneer (and iGem co-founder) Tom Knight with other MIT alumni (Jason Kelly, Reshma Shetty, Barry Canton, and Austin Che), that calls itself "the world's first organism-engineering foundry". A foundry is usually a place where semiconductor chips are manufactured on behalf of companies like Intel. Intel sends the design, and the foundry delivers the electronic chip. Ginkgo Bioworks opened a "foundry" to make living organisms: the customer sends the design, and Ginkgo delivers the organism. So far it has produced synthetic perfumes, cosmetics and foods. Both Zymergen and Ginkgo want to become biotech factories that can produce all sorts of consumer goods in their robotic laboratories. This could revolutionize the chemical industry. Today a lot of materials are obtained via chemical engineering from fossil fuels. Basically, the chemical industry "re-engineers" natural materials (like petroleum) to become artificial materials (like plastic) that are useful to humans Unfortunately this process is not "green": it generates a lot of pollution; and it builds materials that are not "green", i.e. that do not decompose (like plastic). Ginkgo could potentially produce materials that are "green" and using a "green" process.

There are applications of biotech that are not easy to imagine until someone tries them. For example, one classical problem of medicine is how to operate on the brain. Our body is designed to minimize the risks of infection and attacks from the outside, especially inside the skull. The brain is isolated from the blood stream in order to avoid that "attackers" can use the bloodstream to attack the brain. This is a great line of defense, but makes it impossible for doctors to deliver medicines to the brain via the bloodstream. The only way for doctors to help the brain is brain surgery. It would be an improvement if doctors had a way to enter the brain and deliver genes into the nuclei of brain cells in order to reprogram them. For example, doctors could deliver genes that create antibodies inside the brain. In 2015 Ben Deverman's team at Caltech took a harmless virus named AAV9, created millions of genetic variants of it (through the "polymerase chain reaction" or PCR, invented in 1983 by Kary Mullis and still widely used in laboratories), and invented a new technique to test these millions of variants. This was natural selection at lightining speed: their technique selects the variants that are capable of entering the brain brain and of delivering genes to brain cells. Imagine if we could do something like this for all the cases in which today surgery is required to repair a body organ.

This is a field that was born in 1981 when, independently, Martin Evans at Cambridge Univ and Gail Martin at UC San Francisco isolated embryonic stem cells of the mouse. Stem cells are the mothers of all the cells of our body. Once they specialize in a specific job, they cannot be used to make cells of a different kind, but, before they specialize, when they are still "pluripotent", they can develop into all cell types. The stem cells of the embryo are pluripotent. The stem cells of your nose are adult stem cells: they can develop into nose cells, not into liver cells. For more than a decade these studies were limited to other animals, but then scientists started studying the human embryonic stem cells. William Haseltine coined the expression "regenerative medicine" in 1992. It wasn't until 1998, though, that James Thomson at the University of Wisconsin isolated human embryonic stem cells. This step made it possible for scientists to generate all the building blocks of our body in a laboratory. At this point there was enough commercial interest in the possibilities of regenerative medicine that several companies were created all over the world. In retrospect, some of the most influential were: Cellectis (France, 1999), Mesoblast (Australia, 2004), Capricor Therapeutics (Los Angeles, 2005) and Pharmicell (Germany, 2006). In 2004 the state of California launched a California Institute for Regenerative Medicine that has been helping research in the field. Another decade went by, with a lot of controversy about the ethical aspects of stem-cell research, In 2006 (at a meeting of the International Society for Stem Cell Research in Toronto) Shinya Yamanaka of Kyoto University in Japan reported that he had converted adult cells into pluripotent stem cells by simply triggering four genes (now known as the "Yamanaka factors": Oct4, Sox2, C-myc and Klf4). These genes are normally active only in embryos. His seminal paper was titled "Induction of Pluripotent Stem Cells" and that became the term for the technology of genetically reprogramming cells to become pluripotent: "Induced Pluripotent Stem" cells or iPS cells. We don't need to get embryonic stem cells from humans, we can create them in the laboratory. Cellectis immediately licensed Yamanaka's patents.

Converting differentiated cells into stem cells is like turning back the biological clock: Yamanaka's procedure reprograms the DNA in the cell to an embryonic state.

In 2011 Pharmicell got approval for the first stem-cell drug, called "Hearticellgram-AMI". Today Mesoblast, that uses its own proprietary kind of cells, is probably the best known actor in regenerative medicine.

Madeline Lancaster at Cambridge University is using pluripotent human cells to grow three-dimensional tissues ("cerebral organoids") that she uses to model how the human brain develops (see her paper "Cerebral Organoids Model Human Brain Development And Microcephaly", 2013).

This field had its share of scandals. Two rank among the biggest scientific scandals of the last century. In 2004 a Korean scientist, Hwang Woo Suk, announced that he had cloned human embryonic stem-cells, but an investigation by his university found that he was lying. In 2014 a young Japanese scientist, Haruko Obokata, announced that she had discovered a way to turn adult cells back into stem-cells, the so called "stimulus-triggered acquisition of pluripotency" (STAP) cells. Her employer, Riken, investigated and found that it was not true. So we need to be cautious about the announcements that come from stem-cell startups.

But now there is indeed an alternative to Yamanaka's procedure to make stem cells: somatic-cell nuclear transfer (SCNT). In 2007 Shoukhrat Mitalipov at the Oregon Health & Science University created SCNT cells in monkeys, and in 2013 he created human embryonic stem cells from cloned embryos ("Human embryonic stem cells derived by somatic cell nuclear transfer", 2013).

When we mix gene therapy and stem-cell research, we obtain tools that look promising for the regeneration of tissues and body parts. They use different approaches, but this kind of research is going on in several laboratories around the world: Ying Liu at the University of Texas at Houston; Guangbin Xia at the University of Florida; Joshua Hare at the University of Miami; Malin Parmar at Lund University in Sweden; etc. The first major success story of combining stem-cell techniques with gene therapy was announced in 2017 in Italy. Hospitals are familiar with a technique to treat severe burns: it consists in taking cells from the patient's own body (for example, from the groin), growing sheets of healthy skin in the laboratory, and then surgically grafting this "artificial" skin over the wounds. Michele DeLuca's team at Unimore did something similar to cure a terrible skin disease called "junctional epidermolysis bullosa" (JEB), whose genetic causes are now well-known: mutated genes. DeLuca’s team took skin from the patient's groin, whose cells, alas, included the mutated gene, used a retrovirus to insert the right gene into these skin cells, and grew artificial skin in the laboratory, including the stem cells that regenerate the skin. The hospital then surgically grafted the new skin on the body of the patient, and within a month the skin began to regenerate and today it has already covered 80% of the body.

Gene therapy is about making sure that the correct proteins are produced. For example, gene therapy can replace a faulty gene with a functioning one, or introduce a gene that is missing. It can also disable a mutated gene that is causing problems. The ultimate goal is to manipulate genes to restore the function of proteins. It is not trivial to insert genes in an adult body. The operation requires a “vector”, for example a virus: the job of a virus is to infect cells, and we can “hijack” the virus to make sure that it will infect the cell with the desired gene.

A terrible disease called "severe combined immunodeficiency" (SCID) was the first disease to be treated with gene therapy. Children who have SCID (sometimes called "bubble children") are basically without an immune system. In 1990 William French Anderson at the National Institutes of Health (NIH) in Maryland introduced a gene called ADA (Adenosine Deaminase) into the immune cells of a four-year-old girl, Ashanti DeSilva, who was suffering from SCID. It didn't really work well, but that was the first case of gene therapy. The first big success story came in 2009, when an eight-year-old boy, Corey Haas, who was going blind (a form of blindness caused by mutations in the gene RPE65) regained normal vision thanks to gene therapy performed by Jean Bennett at the Children's Hospital of Philadelphia. In 2013 this hospital spun off the company Spark Therapeutics. Research on "bubble children" continued and after more than 20 years Anderson's original intuition started working. In 2013 Bobby Gaspar at the Great Ormond Street Hospital at University College London reported success in treating children suffering from SCID with genetically-engineered stem cells, and in 2014 Donald Kohn at UCLA cured 18 children born with SCID by introducing the ADA gene into their cells. The first gene-therapy treatment approved in Western countries was Alipogene Tiparvovec, marketed since 2012 as Glybera by the Dutch company UniQure. It treats a very rare condition and it costs more than one million dollars, the most expensive medicine in the world. Obviously it was not a commercial success. In fact, UniQure abandoned it in 2017. But a second gene-therapy treatment should soon be approved in Europe: a gene-therapy treatment to treat those "bubble children" born with SCID. It was developed at San Raffaele Institute in Italy and will be marketed by GlaxoSmithKline as Strimvelis.

We increasingly think that T-cells are the secret to improving the immune system. The T-cell (that stands for "thymus cell") is the foot soldier in the army of the immune system: it recognizes and fights the attacking enemies (the "antigens"). This was discovered in 1958 by Jacques Miller at the University of London: he showed that animals without a thymus had a weak immune system. Until then the thymus was considered a useless leftover from evolution: it is actually the organ that keeps us alive on a planet infested with all sorts of enemies for our health. In 1983 Philippa Marrack and John Kappler at the University of Colorado discovered the T-cell receptor (the chemical mechanism of antigen recognition), while, at about the same time, Tak Mak at the University of Toronto and Mark Davis at Stanford University discovered the relevant genes for, respectively, humans and mice: those genes determine how well the T-cells do their job. Now scientists have found a way to reprogram the DNA of a patient’s T cells to secrete a protein called a "chimeric antigen receptor" (CAR) that targets the protein CD19 of tumor cells. This is done in the lab. Then the new cells are inserted in the patient’s body and multiply rapidly.

In 2015 the big success story was Layla Richards, a one-year-old girl who was cured of an "incurable" cancer (acute lymphoblastic leukaemia, the most common form of childhood leukaemia) at Great Ormond Street Hospital . The baby did not have enough of the T-cells that search and destroy the leukaemia cells. Cellectis scientists using TALENS edited T-cells from healthy donors and created T-cells for her. It is relatively easy to inject outside T-cells into the body, but there are two problems: the foreign T-cells don't recognize the body as their own and start killing all sorts of cells in the body that they are supposed tosave, and the body does not recognize the T-cells so it starts fighting them with its own antibodies (like it fights a transplanted organ). Cellectis edited out two genes from the donor's T-cells in order to disable both processes. The result is a T-cell called UCART that is "universal", i.e. that can work in any body: UCART stands for Universal Chimeric Antigene Receptor T-cells. The ones used for leukaemia are UCART19 and will be sold by Servier and Pfizer. In 2016 a second baby condemned by the same cancer was cured by the same UCART19 at the same hospital.

Children with epidermolysis bullosa are condemned to a horrible and painful death: their skin literally disintegrates. Doctors in Holland killed two of these so-called "butterfly children" in order to stop their suffering. This disease is caused by a defective gene that does not produce the protein (type-7 collagen) that holds skin layers together. In 1997 Paul Khavari at a medical center in Palo Alto had published a paper titled "Cutaneous Gene Therapy" that explained how gene therapy could help. Almost twenty years later in 2016 a team at Stanford University, where Khavari now leads a laboratory, has used gene therapy to inject the correct gene into stem cells of the child and to grow healthy skin that can then be grafted on the child's wounds. The gene therapy will be commercialized by a Ohio startup called Abeona Therapeutics (acquired in 2015 by Texas-based PlasmaTech), which is a 2013 spinoff of the Nationwide Children's Hospital in Ohio. The problem is that our body sheds most of its cells every year, so this gene therapy has to be renewed almost every year. It is not a solution, but an improvement over killing the child!

The most interesting T-cells for gene therapy are probably the "chimeric antigen receptors" or CAR-T. The first generation of CAR-T therapy appeared in 1991 and was designed to fight AIDS. It was mainly developed by Brian Seed's group at Harvard Medical School, but similar studies were also under way by Art Weiss' group at UC San Francisco and by Richard Klausner's group at the National Institutes of Health in Maryland. It takes a long time to go from the conference paper to an actual "medicine". This first generation of CAR-T went into clinical trial in 1997. In 2003 a pediatric oncologist, Dario Campana at St Jude Children's Research Hospital in Tennessee, found a new way to make CAR-T cells. Using that method in 2010 Carl June's group at the University of Pennsylvania developed a new, more powerful generation of CAR-T therapy. Cells are taken from a patient's immune cells, genetically modified in a laboratory using a virus, and then reintroduced in the patient's immune system: the genetic modification turns them into hunters and killers of the source of the cancer. In 2011 June's group published the paper "Chimeric Antigen Receptor-modified T Cells in Chronic Lymphoid Leukemia" that was a morale boost for the whole field. (Unfortunately he forgot to credit Campana for his fundamental contribution, which caused some controversy later on). In 2017 the Food and Drug Administration (FDA) approved the first gene therapy for cancer treatment in the USA: a commercial version of Carl June's Tisagenlecleucel or CTL019, sold by Novartis as Kymriah. Unfortunately the price is $475,000! Almost at the same time Gilead offered $12 billion to acquire Kite Pharma, founded in 2009 in Los Angeles by the Israeli oncologist Arie Belldegrun, a firm that had started the clinical trial of its own CAR-T therapy. A few weeks later Gilead also acquired Cell Design Labs, founded by UC San Francisco's synthetic biologist Wendell Lim to commercialize his technology to “programs” T-cells. Wendell Lim and Carl June published a roadmap for the development of next-generation therapeutic cells: "The Principles of Engineering Immune Cells to Treat Cancer" (2017). Juno, Bluebird and Cellectis are other biotech companies working in CAR-T. One problem of course is that it took 26 years from the first generation of CAR-T therapy to the commercial product. Will it take another 26 years to the third generation? The second problem is that today this is an art: a specialized laboratory has to engineer CAR-T-cells for each patient. This is time consuming and very expensive. Later in 2017 the FDA also approved Kite Pharma’s CAR-T therapy Yescarta and Spark Therapeutics’ gene therapy Luxturna for an inherited disease that causes blindness (the outcome of the research by Jean Bennett's team at the Children's Hospital of Philadelphia); three gene therapies in one year.

Before the invention of CAR-T therapy, medicine was about manufacturing molecules that can change what is going on inside the body (basically, biochemistry), CAR-T therapy, instead, is about making living cells, not just molecules. These cells are not made out of primordial matter but by meticulously tweaking the genome of living immune cells to turn them into the equivalent

CAR-T startups of 2017 included: Bluebird in Seattle; Ziopharm, Agios and Mustang Bio in Boston; Sorrento, Poseida and Fate Therapeutics in San Diego; Bellicum in Texas; Cell Medica (formerly Immunocode) and Autolus in London; Calibr, a nonprofit drug discovery division of Scripps Research in San Diego; Celularity, a Celgene spinoff; Cellectis in France; Celyad in Belgium; Cell Design Labs, founded in San Francisco by Wendell Lim and acquired by Gilead; Allogene Therapeutics, founded by former executives of Kite Pharma; Precision BioSciences and Carina in Australia; JW Therapeutics and CARsgen Therapeutics in China; etc. In 2018 Celgene acquired Juno Therapeutics for $9 billion.

In 2020 a Cardiff University team led by Andrew Sewell identified a type of T-cell that could be capable of destroying many different types of cancers including lung, skin, blood, colon, breast, bone, prostate, ovarian, kidney and cervical cancer cells.

Another way to cure cancer is to use "programmable bacteria". Nicholas Arpaia and Tal Danino at Columbia University created genetically-modified microbes that helped the immune system of mice to fight cancer ("Programmable bacteria induce durable tumor regression and systemic antitumor immunity", 2019).

The "miracle cures" due to gene therapy are now multiplying rapidly. In 2015 a seven-year-old Syrian child with epidermolysis bullosa was treated at the Children's Hospital of Ruhr University in Germany by a team led by Michele de Luca of the University of Modena: a piece of the child's skin was surgically removed, its DNA was edited in the lab, and the genetically-modified skin was then grafted back on the child's body ("Regeneration of the Entire Human Epidermis using Transgenic Stem Cells", 2017). In 2015 John Pasi's team at Queen Mary University of London used gene therapy to heal patients whose body cannot stop bleeding (a disease called "haemophilia A") because it doesn't know how to produce the protein "Factor VIII" needed to stop the bleeding. Pasi's team used a virus as the vehicle to deliver the genetic instructions to the liver to start producing Factor VIII. In 2016 Philippe Leboulch's team of the University of Paris performed surgery and DNA editing on a French teenager with a blood disease called "sickle cell disease": first they removed his bone marrow (the organ that makes blood) at Necker Children's Hospital in Paris, then they genetically edited the misfiring DNA in the lab, then they reinserted the bone marrow into the body.

In 2016 Elizabeth Parrish, who has her own startup in Seattle called Bioviva, performed gene therapy on herself and improved her "telomere score", which is higher among young people and low among older people. She managed to recover the equivalent of 20 years of telomere decline. This telomere decline is only one of the many aspects of the overall aging process, but she also performed other gene therapies on herself to reverse other factors. Time will tell if her "younger blood" really helped her live longer, but gene therapy is certainly becoming more real. In 2016 Robert MacLaren of Oxford University led a successful experiment of gene therapy to restore sight in a rare case of eye disease ("choroideremia"): he added a working gene to the cells of the retina in order to compensate for the defective gene that causes the disease. In 2016 Dusan Bogunovic at Mount Sinai in New York showed that people without the gene ISG15 (about 1 in 10 million people) have a stronger immune system that can fight almost all known viruses: maybe removing that gene will solve forever the problem of pandemics and make vaccines a relic of the past. Another success story of gene therapy was announced in 2017 by David Williams at Boston Children's Hospital: his team cured children with a terrible brain disease caused by a single mutated gene. These children soon become unable to eat, walk, hear, see, and they usually die within five years. ALD (adrenoleukodystrophy) strikes about one in 20,000 children.

Until then scientists had edited people’s genes but only altering cells in a laboratory and then returning them to the patient, not directly into the body of the patient. In November 2017 Sangamo Therapeutics in a Oakland hospital (near San Francisco) used ZFN to insert a gene into the body of 44-year-old Brian Madeux that had lacked that gene since birth, precisely into the DNA of his liver cells, hoping to cure him of a rare genetic disease called Hunter syndrome. He accepted to become a living experiment after 26 failed surgeries. The era of "in-vivo genome editing" has begun.

There are also other ways to deal with our genetic mistakes. For example, the gene responsible for the neurodegenerative "Huntington's disease" was discovered in 1993, but no cure exists. In 2015 Sarah Tabrizi's team at University College London began a trial of a "gene-silencing" therapy developed by Isis Pharmaceuticals (later renamed Ionis), a pioneer in "antisense" therapies (it received the first FDA approval in 1992): it doesn't edit DNA but makes sure that a specific gene is silenced. In 2017 Roche acquired the rights on this therapy.

There are also scientists who think that they can just "print" tissues and organs. The idea of applying 3D Printing technology to living tissues sounds appealing, and the first company to try it commercially was Organovo, co-founded in 2007 by Gabor Forgacs of the University of Missouri. Now there are 3D Bioprinting startups in Asia, like Cyfuse (Japan) and Regenovo (Hangzhou, China). But 3D Bioprinting is still mostly studied in universities, particularly at Wake Forest University in North Carolina and by Jeremy Mao at Columbia University. The human body is not as good as other animals at regrowing organs. It is pretty good at healing itself, but not when it comes to tendons, ligaments and cartilage like the meniscus. The meniscus is a body part that millions of people have injured all over the world and usually is not repaired. In 2015 Jeremy Mao showed a machine that can bioprint a human meniscus, and in 2016 scientists at Wake Forest University unveiled a printer designed to print skin cells onto burn wounds.

In 2018 Prellis Biologics, founded by Melanie Matheu in San Francisco in 2016), achieved the 3D-printing of human tissue with capillaries.

3D printing is not only for replacing organs but also for rehearsing operations. This was pioneered in Spain. Before performing heart surgery, the doctors at the Virgen del Rocio Hospital in Sevilla use the BQ's printer Witbox 2 to 3D-print replicas of a patient’s heart in order to study it, and doctors at the Sant Joan de Deu Hospital in Barcelona 3D-print replicas of tumors.

I think that the science of xenotransplantation started in 1993, when David Cooper's team at the University of Pittsburgh discovered that the majority of the human immune reaction targeted a single pig antigen: if we could remove that antigen from the pig, we'd be able to use pig organs on human beings. This discovery lay dormant for two decades because we didn't know how to engineer a pig without that antigen. In 2015 Joseph Tector's team at Indiana University used CRISPR to remove two pig genes. Then they transplanted these generically-modified pig organs into monkeys, one of which has survived for a few months. So they proved that it is possible to remove the genes that we don't want from the DNA of the pig. The other problem is that the DNA of pigs contains sequences that constitute a virus, the "porcine endogenous retrovirus" or PERV, that can infect human cells. In 2015 George Church's team at Harvard University identified 62 copies of this virus in pig cells and used CRISPR to neutralize them.

One company that has been extremely active in this field is United Therapeutics, a Maryland-based firm founded in 1996 by Martine Rothblatt. In 2011 it acquired Virginia-based Revivicor. Revivicor had been spun off in 2003 by PPL Therapeutics, the British firm that cloned "Dolly the Sheep". Revivicor mainly was David Ayares' team, specialists in genetically-modifed pigs. In 2013 Ayares's team transplanted a pig heart in a baboon, which survived almost three years. In 2014 United Therapeutics launched a joint project with San Diego-based startup Synthetic Genomics, founded by (yet again) Craig Venter; and in 2017 the pioneers Tector and Cooper were working at the University of Alabama for a program of xenotransplantation funded by United Therapeutics.

George Church too has co-founded with Luhan Yang eGenesis in Boston to develop genetically-modified pigs to grow organs for human transplant. In 2015 eGenesis edited the pig genome in 62 places at once, and in 2017 it announced the birth of 37 pigs that had been gene-edited in a way for its organs to be accepted by the human immune system.

Pigs and sheep have organs that are roughly the right size for transplantation into humans, so pigs and sheep have become interesting for creating organs that can be transplanted into the people who need them. Using CRISPR, it is now possible to produce pig embryos and sheep embryos that contain human cells. These are called "chimeras", organisms whose cells come from two or more species. In 2016 JuanCarlos Izpisua-Belmonte's team at the Salk Institute created a mouse-rat hybrid: they used CRISPR technology to create mouse embryos without the genes that cause organs to form and then inserted rat stem cells into these embryos. They implanted the embryos into a mouse's uterus and proved that a mouse-rat chimera is possible. The following year, they employed the same procedure to create a human-pig hybrid: a pig embryo was injected with human cells (and the team let it develop up to four weeks). The first scientist to cross species boundaries had been Janet Rossant at Brock University in Canada in 1980, who had produced a chimera combining two mice species (not just the embryos but 38 actual mice); but she had used a complicated surgical procedure ("Interspecific Chimeras in Mammals", 1980). And then Carole Fehilly created a goat-sheep chimera in 1984 at the Institute of Animal Physiology in Cambridge. These chimeras consisted in a random combination of tissue across the body. Izpisua-Belmonte is instead a gene-based procedure that targets the growth of specific organs; i.e. the animal becomes an incubator of transplantable organs. In 2019 his team produced monkey embryos containing human cells. At the same time, the Japanese government gave the green light to Hiromitsu Nakauchi at the University of Tokyo to produce mouse-human chimeras.

The first success story of immunotherapy to fight cancer was announced in 2016 by Steven Rosenberg at the National Cancer Institute. Rosenberg's experiments started in 1985 when he led the first trial on humans of IL-2. IL-2 makes T-cells grow. These are then used to infiltrate the tumor (basically, they are transplanted into the region attacked by the cancer) with a procedure called "adoptive cell transfer". In 1996 he improved his procedure thanks to new information about the genes that represent tumor antigens. In 2006 Rosenberg had tried his procedure on 684 cancer patients and was reporting some success in slowing down or even shrinking the tumor. In 2016 he announced that a woman who had colon cancer was free of cancer. For the first time we have healed someone of cancer using targeted immunotherapy.

One of the leaders in T-cell therapy is now Kite Pharma, founded in Los Angeles in 2009 by Israeli oncologist Arie Belldegrun.

Gene editing can revolutionize immunotherapy in yet another way. There is a substance, messenger RNA, that transmits the instructions of DNA to the place in the cell where proteins are made. In theory, one could reprogram this messenger RNA to make sure that the cell makes the proteins needed to fight a disease. It would be an extension of the immune system. Derrick Rossi, who worked on this problem at Boston Children's Hospital, and Robert Langer of the MIT, probably the most famous researcher in the field of "targeted drug delivery", founded Moderna Therapeutics in 2011 in Boston.

In 2015 Matthew Rabinowitz, now the founder of Natera, and Jay Shendure of the University of Washington sequenced in detail the genomes of two parents undergoing IVF and then inferred the genome sequence of the embryo ("Whole genome prediction for preimplantation genetic diagnosis", 2015).

The ethical problem with these procedures is that the laboratory creates a number of embryos and then only one is kept. In other words, all the other embryos are killed. The scientists pick the healthiest one. Of course, this means that the parents too could pick an embryo and kill all the others. Using IVF and PGD the parents can create dozens of embryos and then pick the ones that (who?) will become tall, or blonde, or more similar to granpa; or someday, when we know more about the relationship between genes and intelligence, they could pick the ones who are more likely to become a scientist, or a painter, or a business man. CRISPR makes it easier and easier to edit genes out. So you can "design" the baby that you want. But it also means that you kill all the embryos that you don't want. All of this because we invented a way to create embryonic stem cells via in vitro fertilization. In fact, someday it will be possible to do all of this without in vitro fertilization. At the end of 2015 a new technique called "in vitro gametogenesis" (IVG) has been tested in mice in Japan by Katsuhiko Hayashi of Kyushu University. This technique allows scientists to create both eggs and sperm in the laboratory. They took skin cells from the mice and created eggs and sperm, then fertilized the eggs to create hundreds of embryos and finally implanted the eggs into a female mouse, and several healthy pups were born. In 2012 Hayashi, working in the team of Mitinori Saitou at Kyoto University, had already discovered how to reprogram skin cells to behave like the cells that generate eggs, and earlier in 2016 Yayoi Obata's team at Tokyo University of Agriculture found out how to turn these cells into eggs without placing them into a female body. A few weeks later Hayashi combined the two procedures and obtained the embryos.

You can use either Shinya Yamanaka's technique or Shoukhrat Mitalipov's technique to reprogram adult cells, such as skin cells, to behave like embryonic stem cells. Then you can use Hayashi's procedure to program these stem cells so they will become eggs or sperm. Then you can fertilize the egg and you get an embryo. In fact, usually the scientists produce many embryos, even hundreds. The goal is to select the "best" one.

This has been done only in mice, but in 2014 Jacob Hanna in Israel has already created the primordial human cells needed to generate human eggs. The next step (to generate the eggs, fertilize them and obtain the embryos) is not far away. Hanna is working with Azim Surani at Cambridge University to complete the cycle.

As Hank Greely has written in his book "The End of Sex and the Future of Human Reproduction" (2016), procedures like IVG will complete the process of separating sexual intercourse and reproduction. In the future a doctor will only need a few cells from a woman's skin and a few cells from a man's skin to create as many embryos as desired. Then these parents will be told the "features" of each embryo and pick their favorite. IVG can even create eggs from the skin cells of a man, and sperm from the skin cells of a woman. IVG will allow lesbians to have babies. It will even allow a woman to be both the father and the mother of a baby, a uniparent (although this presents the same genetic problems of incest). You can take skin cells from a one-year-old baby and create embryos, and the one-year baby will have children. You can even take skin cells from a dead person and make embryos that will become children of a dead person.

Imagine a computer program that allows the parents to play with 100 different embryos: the parents can see a simulation of how each embryo will look like at the ages of 5, 10, 15 ,20... 80. The parents can simulate the life of each embryo and then decide which one they want to be. All the other embryos get thrown in the garbage.

It is just a matter of time before we will be flooded with apps to design our children. To start with, HumanCode, founded in Denver in 2017 by Chris Glode and Ryan Trunck, offers an app that runs on the online platform Helix and costs just $200 and tells future parents how tall their children will be. In 2017 a physicist at Michigan State University, Stephen Hsu (now the founder of Genomic Prediction), unveiled a genetic “predictor” that uses machine learning to estimate height from a person’s DNA.

That day is not far away. In 2013 the first "designer baby" was born, a baby boy named Connor: his parents carefully selected one embryo out of seven grown in the laboratory of Dagan Wells at the University of Oxford. All the other embryos were thrown in the garbage.

I personally don't blame the scientists for "playing God": every doctor plays God when s/he saves the life of someone who is dying of a disease. But society will be truly disrupted when it becomes possible to "design" babies, and it is not clear what the rights and the duties are. Should the parents have the right to choose which embryo will live and kill all the others? And nobody can ask a baby if she wants to be a genetic experiment. someone could take a paper coffee cup that you casually tossed in the trash and turn you into a parent without your knowledge or consent. I am not sure that i would like to be the product of parents who wanted to program my beauty, my intelligence, my skills and maybe even my hobbies.

Editing an embryo is very different from editing a cell in a body: the process of gene editing in an embryo changes every cell in the body that develops from that person, including the sperm or eggs that will pass the changes to future generations. Besides altering the entire body, it will cause a gene drive. Editing changes in the cells of adults are not inherited, so the impact is different. In 2015 Junjiu Huang's team at Sun Yat-sen University in Guangzhou used CRISPR to edit the genes of a human embryo but failed to produce a working embryo: some of the cells did not have the edited gene and some cells had damaged DNA ("CRISPR/Cas9-mediated Gene Editing in Human Tripronuclear Zygotes", 2015). Huang wanted to eliminate the mutated gene causing an inherited disease called beta thalassemia. In 2017 Shoukhrat Mitalipov's team at Oregon Health and Science University succeeded: they removed a gene that causes a hereditary heart defect (hypertrophic cardiomyopathy) from a number of embryos without damaging the rest of the DNA ("Correction of a Pathogenic Gene Mutation in Human Embryo", 2017). The embryos were allowed to develop for only five days, but that is enough to prove the concept. Mitalipov did not actually implant them into a woman: no baby was born. It was meant as a demonstration of what is possible. In November 2018 another Chinese scientist, Jiankui He (in Shenzhen), used CRISPR to disable the gene CCR5 from human embryos to give them immunity from HIV (CCR5 is a protein that the HIV virus uses to gain entry into human cells): this time the scientist implanted the edited embryo a woman and the first gene-edited babies were born .

Regardless of the ethical issues related to experimenting on human babies, He's experiment was intriguing if you relate it to the research that is conducted by Alcino Silva's laboratory at UC Los Angeles. This team has discovered that a CCR5-based therapy regularly used on HIV patients (which basically consists in deletion of the same CCR5 gene edited by He) helps the brain of mice recover motor control after a stroke or after traumatic brain injury. It also seems to make mice generally "smarter", more capable of learning new things. As much as we hate to experiment on human babies, it will be interesting to see if these twins turn out to be "smarter" than the average, and, if so, to see what the public reaction will be: will there be a rush to edit the CCR5 gene from fetuses?

Greely argues that the "killer app" of synthetic biology will be the procedure to avoid rare diseases in babies (there are 6,000 rare diseases, with a chance of about 1% that your child will have one). He then argues that the same killer app will be used to treat infertility, and to give children to lesbian parents. Once it is approved, the social benefits of having healthy babies will become obvious: babies with fewer diseases mean lower costs for health care (especially later in life) and fewer epidemics. Greely thinks that states will provide something like IVG for free because of its public health benefits (exactly like vaccines are mandatory). Imagine a society in which there are no more disabilities. George Church, in his book "Regenesis" (2012), envisioned a factory-style process to produce humans who will be immune to all viruses.

But what willl happen to the others? Will these superhumans, equipped with a superpowerful immune system and with the perfect genes, be willing to live among regular humans who get sick and spread diseases, or with humans who are born phisically handicapped and are a financial burden on the healthy ones? Will these perfect humans be willing to pay taxes that help cure the illnesses of regular humans?

The lesson learned is clear: in general big bureaucracies don't solve problems, they simply use problems to get funding for themselves and pay their salaries. Big corporations have made a lot of money out of the existence of cancer, so let me be cynical and suspect that their motivation to eliminate cancer is not as high as the motivation to develop drugs for people who HAVE cancer.

For the future i believe that an important contribution will come from communities of independent researchers like UC Berkeley's Rosetta@home, the World Community Grid (run by IBM) and Australia's DreamLab, people who are pooling together their computers or smartphones to allow scientists to carry out independent research on cancer. There are also "big data" projects that are collecting billions of data about cancer patients, such as CancerLinQ, launched in 2013 by the American Society of Clinical Oncology, and the Genomics Evidence Neoplasia Information Exchange (GENIE) project, launched in 2015 by the American Association for Cancer Research.

In 2005 a young biologist, Rob Carlson, left the non-profit research laboratory Molecular Sciences Institute in Berkeley (founded by Nobel Prize winner Sydney Brenner and continued his biological experiments at home, and founded his own garage startup, Biodesic. In 2008 Jason Bobe and Mac Cowell founded the DIYbio organization on the East Coast, which is usually considered the beginning of the "do-it-yourself" movement in synthetic biology.

In 2009 four talented young New Yorkers (molecular biologist Ellen Jorgensen, bioengineer Oliver Medvedik, freelance journalist Daniel Grushkin and interdisciplinary artist Nurit Bar-Shai) established the nonprofit organization Genspace to promote biohacking, and the following year opened a shared and public biotech laboratory. In 2009 Angela Kaczmarczyk and others founded the Boston Open Science Lab (BossLab).

Silicon Valley responded with BioCurious, another volunteer-run non-profit organization. It was established in 2010 by a group of young independent biologists (Eri Gentry, Raymond McCauley, Tito Jankowski, Joseph Jackson, Josh Perfetto, Kristina Hathaway). It aimed at providing a "hackerspace for biotech" at its Sunnyvale offices. It marked the rise of a community of worldwide hobbyists taking advantage of public-domain databases of genetic parts. European biohackers had Open Wetlab in Amsterdam and La Paillasse in Paris. In 2010 UCLA organized a symposium titled "Outlaw Biology?" at which biohacker Meredith Patterson delivered the speech "A Biopunk Manifesto". Rob Carlson published a book titled "Biology is Technology" (2010), that became the motto of this movement.

In 2010 two of BioCurious' founders, Tito Jankowski and Josh Perfetto, founded OpenPCR in San Francisco to manufacture a machine that could bring biotech to the desktop, basically a copy machine for DNA. Most genetic applications (such as DNA detection and sequencing) required a machine to perform Polymerase Chain Reactions (PCRs), i.e. to amplify sections of DNA. OpenPCR dramatically lowered the price of these PCR "printers" and made them affordable for individuals. In 2010 Austen Heinz founded Cambrian Genomics in San Francisco to manufacture the first "laser printer for living beings", a machine capable of rapidly and accurately producing DNA. Arcturus BioCloud, founded in 2014 in San Francisco, makes it even easier: it wants to be a virtual bio-foundry for rapid prototyping microorganisms using the cloud to communicate with its users.

In 2003 MIT's professor Tom Knight envisioned a catalog of standardized "biobricks" that synthetic biologists could use to create living organisms. His model clearly mirrored the way the personal-computer industry got started, with hobbyists ordering kits from catalogs advertised in magazines and then assembling the computer in their garage.

In 2003 researchers from MIT, Harvard, and UCSF had unveiled the MIT Registry of Standard Biological Parts, which later folded into iGEM, the International Genetically Engineered Machine. Both iGEM and the BioBricks Foundation were Drew Endy's brainchildren. By 2014 the repository would contain 20,000 biological parts (biobricks). "Open-source" biotech was starting a global grassroots revolution in synthetic biology. Every year the iGEM Jamboree, first held in 2004 in Boston, gathered young bioengineers from all over the world to create new life forms, mostly microbes for useful applications. There would be 2,500 competitors from 32 countries in 2014.

The nonprofit organization AddGene, started in 2004 by MIT's student Melina Fan, helps synthetic biologists share their discoveries, i.e. their biological parts. For example, it ships the DNA material needed for gene-editing to any laboratory that wants to experiment with the CRISPR technique.

It is also interesting that we see the emergence of an "open source" movement in biotech. Sage Bionetworks, a nonprofit organization in Seattle founded by Stephen Friend and Eric Schadt, is clearly inspired by GitHub, the most famous repository for open-source software. Friend got it started in 2009 with a donation of know-how and tools from his old company Merck. Their mission talks about the importance of "open networks of contributors to solve complex scientific problems".

What was still needed was the equivalent of Computer-Aided Design (CAD) for synthetic biology. In 2010 Chris Anderson at UC Berkeley delivered Clotho, an open-source "bioCAD" platform to design organisms. In 2014 Autodesk launched Project Cyborg, a cloud-based platform of design tools for DNA designers.

This worldwide community of biohackers is thriving. As prices get lower and lower, we can expect to see amazing achievements by individuals.

San Francisco-based Enlitic is using deep learning to detect lung cancer in CT images. Lung cancer is one of the hardest forms of cancer to detect, which is why it is usually detected when it's too late.

Arterys, a spinoff of Stanford University's StartX accelerator, has developed an application based on deep learning for General Electric's MRI scanners to detect cardiovascular disorders.

Dell has millions of medical images on its cloud, coming from more than 1,000 health-care providers, and is using the deep learning software from Israel's Zebra Medical Vision to provide automated analysis of those images.

Philips is working with Hitachi on an image analysis system because it has an even bigger repository of images, with more than 135 billion medical images. Its health-care devices (X-ray scans, CT scans and MRI scans) generate more than 2 million medical images per week.

IBM is applying its Watson machine-learning system (and the technology acquired in 2015 from Merge Healthcare) to medical image management, but also to diabetes diagnostics (in partnership with Medtronic, Johnson & Johnson and Apple) and to cancer diagnostics (in partnership with several hospitals), which are being packaged as Watson Genomic Analysis. At the same time IBM is encouraging the collection of patient data via smartphones and their upload to its cloud, where Watson can carry out its learning chores. In 2015 IBM opened in Boston the headquarters of a new division called Watson Health, directed by Deborah DiSanzo, former CEO of Philips Healthcare,

Biogen, a Boston company formed in 1978 by Phillip-Allen Sharp of the MIT, Walter Gilbert of Harvard University and a group of European biologists, is the third largest biotech company in the world. It is planning to generate automated "risk reports" from the 1.6 billion records of genomic data that it owns.

Toronto-based startup A4L (Analytics 4 Life) is using billions of data collected in Canada to develop an algorithm for heart disease.

Unfortunately, i don't think we're replacing radiologists and cardiologists any time soon, but the dream is to store all medical images in a cloud and have the equivalent of Google's or Baidu's "spider robots" crawl this cloud and check each new image for signs of trouble. This will happen automatically, without any need to "request" an image analysis. And new "releases" of the spider robots will automatically re-check all images based on whatever new medical knowledge has become available.

A.I. techniques can also be applied to other aspects of health care. For example, in 2016 AiCure announced a system that uses a smartphone camera, facial recognition software and motion-sensing software to remind a patient of medication and to check that the patient takes the medication.

Let's start with DNA as a computer. DNA is a natural substance for computing because it uses a code and that code obeys strict rules of logic. The pioneer of "DNA computing" was Leonard Adleman at the University of Southern California, who in 1994 created a DNA computer capable of solving one mathematical problem. He found a way to encode a string of data in the sequence of nucleotides and then used the chemical properties of DNA to do the calculation. (For the record, one decade earlier Adleman had coined the expression "computer virus" and one of his students, Fred Cohen, had created the world's first computer virus). But the sensational news came one year later, in 1995, when Richard Lipton at Princeton University showed huge potential for computation in the inherent parallelism inherent of DNA-based computing. This parallelism can help solve some mathematical problems much faster than with electronic computers. A few months later Lipton's students Dan Boneh and Chris Dunworth showed that a DNA computer could break the data encryption system developed by the National Security Agency (NSA) of the USA. That "application" definitely captured the attention of the media. All sorts of mathematicians, computer scientists and biologists became interested in the computational power of DNA. In 1999 computer scientist Mitsunori Ogihara and biologist Animesh Ray at the University of Rochester published the paper "Simulating Boolean Circuits on a DNA Computer" and Ehud Shapiro at the Weizmann Institute in Israel published "A Blueprint for a Biomolecular Computer" (he built the first one in 2001). The first practical DNA computer was unveiled in 2002 and used for gene analysis by Olympus in Japan (a collaboration with Akira Suyama's team of the University of Tokyo), but not much progress was achieved in the following ten years because DNA computers are difficult and expensive to make. Then in 2013 Drew Endy at Stanford unveiled a simple "biocomputer", a computer operating inside a living cell. This computer can only answer "true/false", but the question can be important: it could detect a disease that cannot be detected with today's equipments. The main difference between a biocomputer and an electronic computer is that the biocomputer can interact naturally with the cells of the body. A biocomputer is slow but it can operated in places where electronics cannot be deployed. When biocomputers become practical, we will be able to deploy computation everywhere inside the body. Endy's biocomputers can even communicate with each other: his team worked out a way for transmitting genetic data from a cell to another cell. A sort of Internet is coming to the cells of your body.

Now let's analyze DNA as a nanotech material. Living beings are self-assembling structures. They are not built in a factory: they assemble themselves, cell by cell. The resulting structures are pretty amazing. Think of the human brain: we still cannot design and build in a laboratory anything remotely similar to the human brain, which self-assembles in a mother's womb in 9 months and will keep self-assembling for the rest of the person's life. Nanotech uses two approaches to build new materials: top-down and bottom-up. Top-down is done in many laboratories where the scientists put together molecules or even atoms with painstaking precision, hoping to achieve a stable material. Bottom-up is done when scientists find a structure that will keep growing by itself. Bottom-up is what life does: life is a bottom-up process, i.e. it self-assembles. DNA is an excellent nano-construction material because because we know that it works: it constructs billions of living organisms every single day.

The first man who saw the analogy was probably Nadrian Seeman at New York University. In 1982 he published a paper about constructing 3D structures from DNA that is considered the beginning of DNA Nanotechnology. Nothing happened for 20 years because the machines to "synthesize" (create artificially) DNA structures were very limited. In 2005 Seeman wrote the paper "From genes to machines - DNA nanomechanical devices" realizing that these ideas were becoming feasible. In fact, in 2006 the breakthrough came, although it came from the other side of the country: Paul Rothemund at CalTech showed how DNA molecules can be folded into two-dimensional structures and how DNA structures can be programmed to form larger DNA structures. "DNA origami" (as the technique came to be known) was the cover story of Nature magazine on March 16 of 2006. So the bottom-up approach became popular and John Pelesko published the book "Self Assembly" (2007). Research in DNA nanotech picked up dramatically in 2009, when William Shih's team at Harvard, and Tim Liedl's team at the University of Munich in Germany published techniques to fold DNA that generate self-assembling origamis.

Hiroshi Sugiyama at Kyoto University started working on "DNA origami technology for biomaterials applications" (as his paper of 2012 titles). In 2011 Shawn Douglas of Shih's lab at Harvard founded the BioMod competition to encourage students from all over the world to experiment with DNA origami.

In 2012 two of George Church's students at Harvard University, namely the same Shawn Douglas and Ido Bachelet, developed nanorobots made of DNA with the intention that they could be programmed to target specific cells in the body, for example to seek out cancer cells and program them to self-destruct.

All of this progress was enabled by better machines to "synthesize" DNA (like Agilent machines). It quickly became apparent that these mathematicians and biologists were using DNA to "design" robots the same way that architects use software to design objects. The software to design objects is called CAD (Computer-Aided Design), and the most popular CAD software comes from Autodesk. Similar software was introduced (especially at Harvard) for biologists who want to design DNA robots. CADnano was developed in 2009 by Shih when he was at the Dana-Farber Cancer Institute, and later improved by Church's group and by (not surprisingly) Autodesk. CADnano provides biologists with a way to do what software engineers call "rapid prototyping", except that in this case it is rapid prototyping of three-dimensional DNA origami structures. In 2009 Hao Yan at Arizona State University developed Tiamat, a three-dimensional editing tool for DNA structures. In 2011 Mark Bathe at the MIT developed CanDo (a name that stands for "Computer-aided engineering for DNA origami"), a software that can convert a two-dimensional DNA origami blueprint into a complex three-dimensional structure. In 2016 Chris Voigt's group at the MIT released a programming language called Cello that allows biologists to rapidly design DNA circuits. It is based on Verilog, the language that hardware designers have used for 30 years to design electronic chips. Cello automatically designs the DNA sequence required to implement the DNA circuit. In other words, you can build living cells with this programming language. And so on: these are open-source software for DNA origimi.

Douglas then moved to UC San Francisco in 2012 and Bachelet to Bar-Ilan University in Israel, founding two important schools of DNA origami. In 2013 Bachelet published his own method of making DNA molecules that can be programmed to reach specific places in the body and carry out "special missions" in those places. Basically, the DNA origami was becoming a tiny computer that can travel inside the human body. These DNA computers can perform the same kind of logic operations (zero and one logic) as today's silicon-based computers. Right now they are much less powerful, comparable to the first computers of the 1950s, but at least they are already very tiny.

In 2014 Bachelet in collaboration with Daniel Levner at Harvard inserted such DNA nanocomputers into a living being, a cockroach, and let them travel inside its body. You can envision a day in which these DNA robots will be able to interact with the cells of the body that they "inspect" and also to interact with each other, just like computers can be connected in a communication network. In 2015 Ido Bachelet began the first human trial of DNA nanobots (to fight cancer) and Pfizer invested in his ideas.

Living beings (bodies) are collections of molecular objects that move and interact all the time. Think for example of your blood cells. When they collide, the cells of our body perform sophisticated chemical processes. And these chemical processes, in turn, control the collisions. Bachelet, Douglas and their former colleagues at Harvard are creating artificial kinds of such molecular objects.

The next question, of course, is how much information you can store in the "memory" of these DNA computers. One gram of DNA can hold about 10 to the 14th power Mbytes of data. In 2012 George Church encoded his latest book into DNA. In 2013 Ewan Birney's team at the European Bioinformatics Institute encoded all 154 of Shakespeare's sonnets, an audio recording of Martin Luther King's famous speech "I Have a Dream", and a picture of their office in a string of DNA (a total of 739 kilobytes). In 2015 Sri Kosuri, a member of the Harvard team that had encoded Church's book into DNA, encoded a rock song by the band OK Go into DNA, the first music to be released on DNA. In 2016 Karin Strauss of Microsoft and Luis Ceze of the University of Washington encoded 200 megabytes of data in DNA, including 100 classics of world literature. In 2016 Korean-born Hyunjun Park founded Catalog Technologies in Boston to build a device for DNA storage. These memories are very slow compared with silicon memories, but they can last a lot longer... literally, tens of thousands of years. The problem is the cost of storing data in DNA. Agilent synthesized the DNA for free, but it would normally cost more than $12,000 per megabyte using Agilent's machines that cost millions of dollars. My 16Gbyte flash drive that i keep in my pocket cost $20, and it costs zero to rewrite the data on it. Nonetheless, the storage ability of DNA is impressive: everything that human civilization has produced in writing (50 billion megabytes of text) can be stored in the DNA of the palm of your hand.

You can envision a day when you will have tiny DNA origami robots traveling nonstop around your body and communicating with each other; and maybe they will become so powerful that you can run some A.I. program on them so that they can monitor and interpret what is happening inside your body in real time.

The dangers are of course real, and scientists themselves have done much to protect the science from horrible mistakes; but of course bad people and stupid people abound, and we have to be prepared for the worst every time a new technology is introduced.

I hope that biotech has not forgotten one important lesson from the 1990s. In 1990 William French Anderson carried out the first gene therapy at the National Institutes of Health (NIH). In the following years the expectations for a boom of gene therapy were high. Then in 1999 a teenager, Jesse Gelsinger, died during a clinical trial of gene therapy at the University of Pennsylvania. This tragic event killed gene therapy for two decades. All it takes is one mistake and an entire field comes to a halt.

The other danger is that the biohackers may create something that will not be easily "undone". The "undo" command doesn't exist in biotech. In 2014 George Church, the influential Harvard bioengineer, and Kenneth Oye, a political scientist at the MIT, published an article in Science magazine that gene-editing techniques and gene-drive techniques are too dangerous when they leave the scientific laboratory. Maybe we should enact a new law mandating that biotech companies introduce a new product only when it is clear what the "undo" process is. If they don't know how to "undo" something that they are doing in the laboratory, then they should keep it in the laboratory.

I also think that philosophers and psychologists are not spending enough time thinking through the fundamental issue of "who am i?" We haven't really spent much time thinking of what happens to "me" when you change one of my genes or reprogram one of my cells. When i teach the class on Neuroscience, i ask my students whether they would be willing to replace their skin. What i really want to know is how much they care for their brain, but before i ask them complicated questions about brain operations, i test them by asking the same questions about something as simple as the skin. The human skin is not a great material: it gets cut and burned very easily. How about i replace your skin with some metallic material like stainless steel that does not break and does not burn? You don't have to worry about scratches, bleeding, bruises, cuts, and burns: happy? And this new material even insulates you from cold weather. Are you willing to replace your skin with this new skin? After thinking about it, most students reply "no". The psychology behind that "no" is simple: "I" (note the "I") prefer to stick to "my" skin (note the "my") because that is "me". If you change my skin, i am not sure that i am still "me". My instinct tells me that i become a cyborg, some strange kind of living being, probably more efficient, but i lose my identity. Now the bigger question is: can i improve your brain to make you smarter? That is a big question because that "improved" brain would be someone else's brain: you would become another person, a more intelligent person, but not "you". I know that i am not the most intelligent person in the world, and maybe i am the most stupid person in the world, but that's "me" and if you change "me" to some other brain, it is just like killing me. Thanks, but i don't want to die, so i'll keep my stupid brain. So... the same question about genes and cells. We are not spending enough time discussing what happens to "me" when you change one of my genes or when you change the program in one of my cells. The goal is to make me healthier, but are you changing "me"? You are definitely changing one of the organs of my body, and therefore my body: is the result still "me"? There are profound philosophical issues related to meddling with someone's genome. Just like we wouldn't like a brain transplant (someone else's brain in my body is not "me"), maybe we shouldn't want a genome transplant (an operation that alters "my" genome).

But the biggest danger, perhaps, is to feel too confident about our understanding of genetics. For example, we all know that the DNA has the structure of a double helix. Our genome is expressed as a sequence of base letters, and this sequence is physically encoded in a double helix. But that is true only in cells that are at rest, which mostly happens when they are dead. In living cells usually the shape is more complex because the helix twists and loops in irregular geometric ways. When cellular biologists discovered techniques to read the sequence of the base letters (such as TALEN and CRISPR techniques), we entered the age of gene-sequencing, and scientists largely stopped studying the meaning of the changing shapes of the double helix. In information terms, we contented ourselves with studying the lower-order structure of DNA (the double helix) and ignored DNA's higher-order structure, which is actually what biologists find in most living processes. The dogma of lower-order cellular biology is that some special proteins attach to DNA and trigger gene replication or gene expression, and that's the essence of cellular life. But in reality those same processes of replication and expression can take place even without the active role of proteins. When the double helix swings, it can achieve the same effects. At a higher level it becomes apparent that DNA and proteins influences each other. There is a whole field of "DNA topology" that is still largely unexplored.

There are many mysteries in the way the genome works. Craig Venter's team simplified as much as possible the genome of a bacterium and in 2016 published the smallest set of genes that can still be a living organism: 473 genes. If you remove any of those 473 genes, the organism cannot survive. The problem is that we don't understand the function of more than 150 of those genes.

In 2016 Stephen Friend of Seattle-based nonprofit organization Sage Bionetworks and Eric Schadt of Mount Sinai published a report that demonstrates how little know about the human genome. There are millions of healthy people who, according to their genes, should be very unhealthy. These are people whose genomes contain genetic mistakes which are known to cause devastating illnesses. But they are perfectly healthy. Jillian Banfield's team at UC Berkeley is using the genomes of animals to re-design the tree of living beings, and it is not the one we knew.

There are even bigger mysteries in the way the genome translates into a living being. The human genome contains 25,000 genes, but rice contains 50,000. So a grain of rice is more complex than me?

In 2016 Steven McCarroll's team at the Broad Institute in Boston announced the discovery of the genes involved in schizophrenia, and a few months later Serena Nik-Zainal's team at the Sanger Institute in England published the genes involved in breast cancer. We have to be very careful to use these data. There have been many cases when society trusted science too soon. For example, in the 1920s eugenics was a very popular scientific topic in the universities of the USA: a few years later eugenics was used by Hitler to justify the extermination of Jews and by Japan to justify mass murder and rape in China. For example, psychiatry was very popular in the USA until recently, and psychiatrists came to dominate many psychology departments in the main universities, but most of Freud's theory has been proven wrong by modern neuroscience.

In conclusion, i hope that biotechnicians realize how little we know about life. After all, this is a very young science. It has only been 60 years since we discovered the double helix of the DNA.

But there is also the opposite risk: that society will be too slow to accept the progress in biotechnology. The US agency in charge of approving new drugs, the FDA, does not "scale up": it cannot approve 1,000 or 2,000 new organisms per year. It takes several years to analyze a new biological product, and only 40-50 are approved any year. On one hand, the public is scared and wants to be protected by strict and tough regulations. On the other hand, biotechnology can deliver a lot more than it does today and these regulations hamper progress that could save the lives of millions of people and these regulations also make it very expensive to conduct research and development. There is zero tolerance for errors in biotech because governments are scared that an error could kill many people, but sometimes the result of this zero-tolerance policy is to let millions of people die of diseases that could be cured and another result is to make all drugs very expensive. The country that figures out how to reform regulations and make it easier and cheaper to introduce new organisms in the world will have a huge advantage over the rest of the world.

Then I am also afraid of the pharmaceutical industry, an industry that missed the manufacturing revolution of the 20th century. "Continuous" manufacturing has become the norm in almost all manufacturing industries since Oliver Evans built his flour mill more than 200 years ago. The pharmaceutical industry is the exception: it is still living in the world of "batch" manufacturing. It can take one month to manufacture a drug that could be made in two days. In 2007 Novartis created the Center for Continuous Manufacturing at the MIT, which in 2012 spawned the startup Continuous, and in 2016 the MIT showed the first portable machine that can make a drug starting with the raw ingredients. The future of "pharma" could be "portable drug making".

My problem is that i don't trust Big Pharma. The scientists discover a new cure. The startups rush to develop a product. But then it is usually "Big Pharma" that brings that product to the market. Can we trust Big Pharma? Absolutely not. Their goal is to make money, not to cure everybody. If everybody were healthy, they would be out of business. It has happened more than once that Big Pharma cured a disease by creating a bigger one. The most recent scandal is about painkillers and was discovered in 2017 by the Washington Post: Big Pharma bribed politicians and used dubious organizations like the American Pain Foundation to make sure that highly-addictive treatments based on opium would be approved. In 2016 the number of people who died of opioid addiction in the USA was 64,000. 75% of opioid addicts in the USA got addicted by legal medicines, not by heroin. And you cannot tell me that doctors were not accomplices in this scam. Doctors knew that those treatments were dangerous and they still got thousands of people addicted to opioids. In 2016 the doctors in the USA prescribed 236 million opioid treatments. An investigation even revealed that powerful painkillers meant for cancer patients were given to patients who didn't have cancer at all. How can we trust Big Pharma with much more risky treatments like gene therapy?

Genomic databases are promising for "deep learning" because neural networks work well when there are big data. Schizophrenia is a heritable brain disease, but the genes responsible for it have been challenging to detect. In 2016 Steven McCarroll's team at the Broad Institute in Boston announced the discovery of the genes involved in schizophrenia. After analyzing the genomes of more than 64,000 people, the team determined that people who have the overactive forms of gene C4 are more likely to be schizophrenic. Suddenly everything made sense. It was known that the brains of schizophrenic people have a thinner cerebral cortex with fewer synapses and one could guess that this was due to excessive pruning of synapses. The gene C4 is expressed by human neurons and causes that kind of pruning. The higher the levels of C4 activity are, the greater a person's risk of developing schizophrenia is. A few months later Serena Nik-Zainal's team at the Sanger Institute in Britain discovered five additional genes involved in breast cancer after analyzing more than 500 genomes of people with breast cancer. In 2017 the world's largest study on the genetics of breast cancer, done by more than 300 research groups worldwide by analyzing the genomes of over 275,000 women, discovered 72 new gene variants that are likely to cause the disease (the enormous team was directed by Georgia Chenevix-Trench of the Queensland Institute of Medical Research in Australia). In 2017 Insilico's lab in Moscow used a neural network and discovered new molecules that can help fight cancer ("The Cornucopia of Meaningful Leads", 2017).

There are also databases of protein properties like Kamil Tamiola's Peptone (2016, Holland). Protein design and engineering is not an exact science because the relationship among sequence, structure and function in proteins is not fully understood.

There are so many databases that one can start thinking of full-body virtual simulation. Firms like Insilico Medicine (2014, Baltimore) and iCarbonX (2015, Shenzhen) seem to be thinking in that direction: create a complete simulation of the patient's body, at many different levels of description, from the genetic to the physiological, and then use A.I. to analyze the body's health. Government research centers like the Virtual Physiological Human Inst (EU) and Insigneo (Britain) have been working on full-body simulation for several years but need to add the genetic level to their simulations.

The first kind of help can come from the hot topic in A.I., deep learning, that can be used to "mine" the vast databases that are being created in different areas of biophysics.

Sophia Genetics, founded in 2011 in Switzerland by Jurgi Camblong is using AI to accelerate genetic diagnostics for oncology, hereditary cancer, metabolic disorders, pediatrics and cardiology via DNA sequencing. The data analysis takes a few days with their platform instead than several months (Zhenyu Xu is the brain behind their A.I. system). DNAlytics, founded in 2012 in Belgium by Thibault Helleputte, has a product called RheumaKit that analyzes biological and clinical data to diagnose and treat arthritis disease. Innoplexus, founded in 2011 in Germany by two Indian graduates (Gaurav Tripathi and Gunjan Bhardwaj), automates the collection, curation, aggregation and analysis of data from thousands of sources, optimizing the whole process of drug development from synthesis to approval.

The most interesting applications of deep learning to biotech are in the field of “omics” research (genomics, transcriptomics, epigenomics, proteomics, metabolomics, etc). There are at least three kinds of applications: protein structure prediction, gene expression regulation, and protein classification.

The primary structure of a protein is simply the sequence of amino acids. The secondary structure has been known since Linus Pauling theorized about it in 1951. The tertiary structure is the three-dimensional structure that pharmaceutical firms are interested in. Predicting the structure of proteins is important because predictions of tertiary structure are increasingly demanded due to the rapid discovery of proteins; but tertiary structure prediction depends on secondary structure prediction, and that is not trivial. There are two ways to make the prediction, known as "three states and "eight states". The first scientists to apply neural networks to this problem were probably Ning Qian and Terrence Sejnowski in 1987 (Sejnowski was an influential figure in A.I.). In 1999 David Jones developed the two-stage neural network method PSIPRED. The accuracy of three-state secondary-structure prediction has increased over the years: 69.7% by a method called PHD in 1993; 76.5% by PSIPRED in 1999; 80% by Structural Property prediction with Integrated Neural nEtwork (SPINE) in 2007 (Ofer Dor and Yaoqi Zhou at the State University of New York in Buffalo); 82% by Structural Property prediction with Integrated DEep neuRal network 2 (SPIDER2) in 2015 (Yaoqi Zhou's team, now at Griffith University in Australia); and finally 84% by Deep Convolution Neural Field network (DeepCNF).

Meanwhile, in 2014 Jian Zhou and Olga Troyanskaya at Princeton University developed ICML2014, a deep-learning approach to 8-state prediction. The following year Sheng Wang and Jian Peng (University of Chicago) in collaboration with Toyota Technological Institute created DeepCNF (Deep Convolutional Neural Fields) for both 3-state and 8-state SS prediction, using a number of datasets (CullPDB53 of 6125 proteins, CB513 of 513 proteins, CASP1054 and CASP1155 datasets containing 123 and 105 domain sequences, respectively, and CAMEO). DeepCNF pushed the 8-state accuracy to beyond 70%.

Deep learning has also been used to explore various facets of gene expression regulation. For example, DNA- and RNA-binding proteins play a central role in gene regulation and knowing their sequence is important to explain the regulatory processes and for investigating the genetic causes of diseases. In 2015 Babak Alipanahi at the University of Toronto (the birthplace of deep learning) developed DeepBind to model this process and then founded Deep Genomics. Another example is NIH’s Library of Integrated Network-Based Cellulanatures or LINCS: to save costs, it profiles the expression of only about 1000 landmark genes from the Connectivity Map (CMap) project and then infers the expression of the remaining target genes (whose gene expression is known to be correlated to the landmark genes) via linear regression. This way the LINCS program had generated about 1.3 million gene expression profiles, but linear regression is not a satisfactory method. Xiaohui Xie (in 2016 at UC Irvine) built D-GEX to infer the expression of target genes from the expression of the “landmark” genes using the Gene Expression Omnibus dataset (111,000 expression profiles). Wyeth Wasserman (in 2016 at the University of British Columbia) tackled another problem with his DECRES: identifying enhancer and promoter regions in the human genome. Regulation depends on promoters and enhancers, but detecting the locations of promoters and enhancers (a focus of bioinformatics for twenty years) is not trivial. He used the Encyclopedia of DNA Elements (ENCODE) and the Functional Annotation of the Mammalian Genome (FANTOM), two large databases. DeepTarget was built at Seoul National University in Korea in 2016) for predicting microRNA-mRNA pairs, another crucial information to understand gene regulation. MicroRNAs (miRNAs) are short sequences of ribonucleic acids that control the expression of target messenger RNAs (mRNAs) by binding them. Noncoding variants are statistically associated with human disease, but determining their mechanism is not trivial: in 2015 Jian Zhou and Olga Troyanskaya at Princeton developed DeepSEA to predict the functional effects of noncoding variants from DNA sequence. David Kelley (in 2016 at Harvard) built the open-source deep-learning framework Basset that can be used to learn functional activities of DNA sequences, annotate every mutation in the genome with its influence, etc. RNA splicing is a critical step in gene expression whose disruption contributes to many diseases, including cancers and neurological disorders; and tens of thousands of genetic variants may alter splicing. So Hui Xiong, Babak Alipanahi and Leo Lee (in 2015 at the University of Toronto) built a system to predict the splicing activity of individual exons. In particular, their system provided insight into the genetic basis of spinal muscular atrophy, hereditary nonpolyposis, colorectal cancer and autism.

Then there's protein classification. Ehsaneddin Asgari and Mohammad Mofrad (in 2015 at UC Berkeley) conceived ProtVec, a general representation that can be employed in a wide array of bioinformatics research such as family classification, protein visualization, structure prediction, protein-protein interaction prediction, etc.

In 2012 Hinton's student George Dahl at the University of Toronto won a Merck contest with a multitask deep neural networks for classifying Quantitative Structure-Activity/Property Relationship (QSAR/QSPR). QSAR studies are important to predict the drug properties called ADME (absorption, distribution, metabolism and excretion). In 2015 Vijay Pandey's group at Stanford designed a similarly multitask network for drug discovery that even improves as additional tasks and data are added.

A.I. can also help parents "design" their babies with PGD and IVG, like Stephen Hsu did in 2017. In the same year Denver-based HumanCode introduced a smartphone app to predict your babies’ height (i don't think this one used any A.I. but it's representative of the apps that could be available soon).

But there is one branch of A.I. that has been a little forgotten in the age of "deep learning". Starting in the 1970s, several A.I. laboratories studied how to automate the process of theory formation, i.e. the job of the scientist. This branch of A.I. was motivated by the goal of creating truly intelligent machines, machines that not only recognize objects or words, but can also come up with new ideas. This theoretical research could be very important to help scientists formulate new theories. The field that can benefit most from such help is the biomedical field. It is extremely expensive and difficult to discover new drugs to treat diseases. The amount of information on each disease is colossal and no human being can analyze all of them in one lifetime. In 2016 more than 1.2 million papers were published in biomedical journals, on top of the 25 million already in print; but on average a scientist reads only about 264 papers per year. For example, more than 70,000 papers have been published on the tumor suppressor p53. Today the average cost of drug discovery in the USA is $2.5 billion and, on average, it takes 12 years of trials for each new drug before it is approved. If Artificial Intelligence software and robots can speed up discovery of new drugs, the biomedical field will be revolutionized.

The basic kind of help can come from better search engines to find the most relevant texts from the millions of papers and patents. Sparrho (developedin 2013 in Britain) is probably the most used and in 2017 the Chan-Zuckerberg Foundation acquired Meta, a science search engine started in 2010 in Toronto by Sam and Amy Molyneux.

But the real help would come from automating and accelerating the process of drug discovery. In a 2003 talk at the American Association for Artificial Intelligence, Lawrence Hunter of the University of Colorado proposed a revised Turing Test that would use the publication of a paper in a peer-reviewed scientific journal as a better test for human-level intelligence. In the same year (2003) William Agresti of John Hopkins University published the manifesto "Discovery informatics". In October 2016 Hiroaki Kitano of Sony, president of the Systems Biology Institute (and founder of the Robocup competition), delivered a talk titled "Artificial Intelligence to Win the Nobel Prize and Beyond" in which he proposed a new grand challenge for AI: to develop an A.I. system capable of scientific research in the biomedical sciences that can discover something worthy of the Nobel Prize.

Rein Vos at Maastricht University published the chapter "The Enigma of Drug Discovery" in the book "Drugs Looking for Diseases" (1991) in which he showed that the process of drug discovery can be modeled, and therefore automated.

I don't expect the machine to invent a new scientific theory like Quantum Mechanics or Relativity, but at least to design better molecules, enzymes, and peptides that can be used for the treatment of diseases.

Few people realize that the largest and most valuable AI company in Europe is Benevolent, founded in 2013 in Britain (as Stratified Medical) by Ken Mulvany (who had successfully sold his previous biotech startup Proximagen), that is using Artificial Intelligence to accelerate drug discovery. In 2017 Benevolent was a "unicorn" (worth more than $1 billion) although it never published scientific papers.

Atomwise, founded as Chematria in 2012 in San Francisco by former scientists of the University of Toronto, including Abraham Heifets (author of the public database of patented chemical structures SCRIPDB and of the protein analysis tool LigAlign), has developed a deep-learning system, AtomNet, that "understands" complex concepts of organic chemistry by breaking them down into smaller and smaller concepts.

In 2017 British-based Exscientia, a 2012 spinoff of the University of Dundee, started a collaboration with GlaxoSmithKline and Sanofi. Exscientia's rapid-prototyping platform automates drug design via an expert system that is equipped with a repertory of best practices acquired from experts of the sector. This system can design millions of novel compounds and calculate for each how effective it is likely to be for a specific project. Then it can select the best ones for experiments. Exscientia specialized in the design of molecules that target more than one place because most diseases are actually a network of diseases, so they require therapies that target multiple places.

Berg, started in 2006 in Boston by Niven Narain and funded by billionaire Carl Berg, uses A.I. to analyze genomic and clinical data about a disease and then infers the network of protein interactions that cause the disease. In 2017 AstraZeneca announced a collaboration with Berg to evaluate new treatments for Parkinson’s disease and other neurological disorders. Berg collects samples from diseased and healthy patients, analyze their genome, proteome, lipidome and metabolome, and then compares them with patient clinical information.

Stanford-spinoff TwoXAR, cofounded in 2014 by Andrew Radin and Andrew Radin (yes, two people with the same name), identified a potential drug for liver cancer in just four months by screening 25,000 potential candidates in a joint project with Stanford (the only treatment approved by the FDA took five years to develop).

Recursion, founded in 2013 in Utah, uses computer vision to look at cells and analyze more than 1,000 features to determine whether a sick cell is being "cured" by the compounds that it massively produces. It committed to discovering 100 disease treatments in ten years.

Numerate, founded in 2007 in the Bay Area by Brandon Allgood and Nigel Duffy (later hired by Sentient), has developed learning algorithms to help research projects that typically have few data, noisy data and biased data. Its goal is to rapidly produce drug candidates for diseases such as cancer, hepatitis C or HIV.

Insilico, founded by Alex Zhavoronkov in 2014 in Baltimore, looks at drugs that are already safe to use and see if they can be re-purposed for other uses.

Desktop Genetics, formed in 2012 by three Cambridge University scientists, is applying deep learning to CRISPR gene editing. Their systems (GuideBook, DeskGen, AutoClone) guide the scientist throughout the entire process, from the design stage (selecting proper sgRNA molecules) all the way to analyzing the results of the experiment.

BioAge Labs, founded in 2015 in Berkeley by Kristen Fortney, wants to accelerate a the discovery of a specific kind of drugs, the once for longevity.

PathAI was founded in 2016 in Boston by Aditya Khosla and Andrew Beck (who at Stanford buit one of the earliest A.I. systems for cancer pathology) to provide end-to-end automation for "scalable" discovery. It is working with Philips on deep-learning system to diagnose breast cancer.

In 2018 Daphne Koller, formerly a cofounder of Coursera and a scientist at Calico, started Insitro in South San Francisco to accelerate drug discovery with deep learning, and signed a deal with Gilead.

The worm called planaria has intrigued biologists for more than a century because it is amazingly good at regenerating its organs: from just 1/279th of its body it can regenerate into a full worm within one weeks. If we discover how it does it, we could unlock the secret to regenerative medicine (i.e., make human organs grow again). In 2015 Michael Levin at Tufts University presented the first model of regeneration discovered by a non-human intelligence and the first comprehensive model of planarian regeneration, which had eluded human scientists for over 100 years. The machine reverse-engineered the regeneration mechanism of planaria—the small worms whose extraordinary power to regrow body parts has made them a research model in human regenerative medicine. In 2016 IBM's Watson discovered five new genes linked to ALS (amyotrophic lateral sclerosis or "Lou Gehrig's disease") at Robert Bowser's lab at the Barrow Neurological Institute in Arizona: it analyzed all published literature related to this disease and ranked genes based on the probability that they would be responsible for the proteins known to be associated with the disease. Barrow's team verified that eight of the top ten genes chosen by Watson are indeed associated with the disease, and five of them were previously not suspected.

In 2017 GlaxoSmithKline, the Lawrence Livermore Lab and the National Cancer Institute formed the Accelerating Therapies for Opportunities in Medicine (ATOM) consortium with the goal to transform drug discovery from the slow, sequential and failure-prone process that is today into a rapid and accurate process (from target to patient-ready in less than one year).

In a sense, A.I. could mark the end of the mass-produced drug because it could discover the specific drug that works best for your case, which may be useful only in your specific case. Mendel.ai, founded in 2016 in San Francisco by Karim Galil, Ruchi Deshpande (later hired by Adobe) and Wael Salloum, want to provide customize treatments to cancer patients based on the latest published data. It employs a natural-language processing system to analyze unstructured text of medical publications and then a neural network to compare the content with a patient's medical record. They are working with the Comprehensive Blood & Cancer Center of California to provide the doctors with instantaneous information about which clinical trials might be good for their patients.

Today, according to the Tufts Center for the Study of Drug Development, it takes on average 12 years and costs about $2.6 billion to introduce a new drug on the market.

In late 2016 the FDA approved the first deep-learning system for medical use. It's a cloud-based diagnostic system for heart conditions made by Arterys, a spinoff of Stanford University's StartX accelerator (founded in 2012 by Fabien Beckers, Albert Hsiao, Shreyas Vasanawala and John Axerio-Cilies, and now based in San Francisco). This deep network, sold in conjunction with General Electric's MRI scanners as ViosWorks, takes an average of 15 seconds to produce a result for one case which would normally take an hour by a professional cardiologist. Not quite "biotech", but an important first step towards accepting A.I. as a "cure".

People used to starve to death for lack of food or die of cold for lack of warm housing. We solve the problem of starvation with the agricultural revolution. We solve the problem of housing with the industrial revolution. The 2016 report by the World Health Organization (WHO) showed that cases of diabetes have nearly quadrupled since 1980: more than 400 million people have diabetes now. And each year 3.7 million people around the world die of diabetes. If this trend continues, soon 1 in 10 people of the planet will have diabetes. The same organization estimates that cases of cancer (that already kills 8 million people each year) will increase by about 70% over the next 2 decades. There is still one big problem to solve that neither the agricultural nor the industrial revolutions solved: diseases. That requires the "biological revolution" that is going on today. Centuries from now the historians will write books about the biological revolution of the 21st century.

Some final thoughts about the "longevity economy" (from the title of a book by Joseph Coughlin, the founding director of the MIT AgeLab). The best birth control has been the combination of education and wealth: when people live longer and wealthier lives, they make fewer children. But solving the problem of poverty and disease has created another problem, in fact the biggest disruptive force of all: disruptive demographics. The world will have more and more old people, fewer and fewer young people. By 2020 the population of people over 60 will control 30% of global spending, the equivalent of China's entire GDP ($9 trillion). Every business will be disrupted: retirement plans, of course, but also real estate, tourism, smart cities, international affairs, ... Worse: what will people do when they are 80, 90, 100? We don't have rituals, habits, sports, etc for such old people. They are supposed to sit in front of the TV set and wait to die. That is an incredible stupid way to deal with a growing sector of the population.

Longevity science is multidisciplinary not only because it involves biology and computation but also because it is a truly new field that needs new thinking to deal with an unprecedented situation. People want longer lives and better lives. This can be achieve with "longevity technology" that must start from the day that the baby is born, planning the baby's behavior to maximize the chances of living as long as possible as healthy as possible But we also need to think about what these babies will do when they reach the age of 80, 90, 100, maybe 120, 130, 140...

In many societies you are considered old, and sometimes forced to retire, when you turn 65. But life already extends way beyond 65. Society needs to keep "old" people active. And they need to become self-sufficient: we don't have enough babies to support so many old people, and today's children are no return on investment because they tend to move away from the parents. Society needs to keep "old" people productive. There are two problems: 1. Today tech is developed by young people for young people; 2. Old people neither invent nor adopt new technologies. Who can develop tech for old people?

"Assisted living" technology actually exists already. Think of hearing aids or contact lenses. They are very common and help old people do things that a century ago were reserved to young people. But also think of the microwave oven, that allows people to cook a meal in a few minutes, or of the garage opener, that allows people to open their garage without using any strength. These are technologies that everybody likes, but turn out to benefit more the older than the younger.

There are many technologies that can be used to create new "assisted living" devices: the Internet of Things can make all appliances in the house easier to control; the sharing economy can help people move around; virtual reality can help create virtual tourism for those who cannot travel long distance anymore; or space technology can create supersonic travel for people who cannot spend many hours on an airplane. We already have wearables for health monitoring and they will become more sophisticated and interconnected. We can have robots that will help old people remember when to take their pills, that can deliver packages and pick up packages, and maybe that can even go shopping. Social media are already helping people of all ages find friends and share experiences. Cloud computing, big data and A.I. will help analyze health conditions and prescribe activities, medicines, diets. Chatbots will allow people to shout to appliances instead of having to type on tiny keyboards. And maybe 3D printing will allow people to manufacture the object that fits perfectly their mobility.

And what else? We will need old people to innovate and to create their own economy. That is one of the big challenges of the 21st century. The 21st century will end with a much older society than we are used to.

Boston biotech startups of 2018 included: Foundation Medicine, founded in 2010 as a spinoff of the Broad Institute and basically acquired in 2015 by Roche (cancer genomics); Rubius Therapeutics, founded in 2013 (red-cell therapeutics); Semma Therapeutics, founded by Doug Melton in 2014 (cell therapy for Type 1 diabetes); Magenta Therapeutics(2015, transplant medicine); BlueRock Therapeutics (2016, engineered-cell therapy); Relay Therapeutics (2016, therapeutics involving protein motion); Shepherd Therapeutics (2016, rare cancer therapeutics); Tango Therapeutics (2017, cancer therapeutics); etc.

Some old ideas are finally becoming commercially available. It was known since the 1960s that there is free-floating DNA in the blood. In 1997 Dennis Lo at the University of Hong Kong even detected fetal DNA in the plasma of a pregnant mother. In 2011 Hong Kong clinics introduced the first "noninvasive" prenatal genetic testing. Indirectly, that research realized that one can test DNA anomalies caused by cancer simply by looking into a person's blood because cancers shed DNA into the bloodstream. In the age of DNA sequencers, it has become technically feasible to develop a cheap mass-market screening test for early detection of cancer. Some of the most funded startups of 2016-17 are working on blood tests to detect cancer, so called "liquid biopsy". Guardant Health, started in Redwood City in 2012 by Illumina executives AmirAli Talasaz and Helmy Eltoukhy, developed a blood test that looked at 73 genes for a variety of cancers. Freenome, founded in 2014 by Gabe Otte, is essentially a software company developing tests for several types of cancer. Grail, founded in 2016 in Menlo Park by former Google executive Jeff Huber, raised more than $1 billion in one year (possibly the most heavily funded biotech startups ever), and in 2017 teamed up with Cirina, the startup created by Dennis Lo in 2014 in South San Francisco. They were joined in 2017 by Sunnyvale-based Apostle, started by former Gilead executive David Dongliang Ge.

Another popular specialty is cancer immunotherapy, well represented by Arcus Biosciences, founded in Hayward in 2015 by Juan Jaen and former Amgen executive Terry Rosen (both previously cofounders of Flexus Biosciences), and Gritstone Oncology, founded in Emeryville in 2015 by Andrew Allen and Mark Cobbold.

23andMe remains the world leader in personalized genetic testing and others personal genomics startups have to find their niche. San Francisco-based uBiome, founded in 2012 by alumni of the the California Institute for Quantitative Biosciences (Zachary Apte and Jessica Richman), sequences the microbiome, while Pleasanton-based 10X Genomics, founded in 2012 by Serge Saxonov (the chief architect of 23andMe's technology), uses software to learn more from Illumina’s sequencers and in 2018 acquired Stanford spinoff Epinomics, a startup in the new field of epigenomics.

Epigenomics is the study of how genes switch on and off. In a sense, it is wrong to view the genome as the program of a body. The genome is a static set of data. Something activates functions in the body based on those data, i.e. regulates those genes and makes them "express" themselves. Those are the instructions, the real "program" of the body. The pioneer in epigenomics was Cambridge Epigenetix, a 2012 Cambridge University spinoff.

The biggest biotech acquisition in the Bay Area until 2016 was the one of Stemcentrx, acquired by Chicago-based AbbVie for more than $10 billion. The startup was active in another thriving field. In 1997 John Dick and Dominique Bonnet at the University of Toronto discovered that leukemia is caused by a cancer stem cell, and ten years later John Dick was able to convert normal human blood cells into leukemia stem cells. These experiments hinted that most solid tumors may rely on a subpopulation of tumor-initiating cells that came to be known as "tumor stem cells". Then several startups tried to develop an "antibody drug conjugate" or ADC to fight such tumor stem cells. Cellerant, founded in San Carlos in 2003 by Stanford scientist Irving Weissman, developed an ADC to target leukemic cancer stem cells. OncoMed, founded in 2004 in Mountain View, tried to commercialize the 2003 discovery by Michael Clarke and Max Wicha at the University of Michigan that breast cancer too is caused by tumor stem cells. Similarly, Stemcentrx, founded in South San Francisco in 2008 by Weissman's student Scott Dylla, worked on an ADC to attack tumor cells.

See also my review of Helen Pilcher's book "Life Changing" (2020) that contains more updated information on cloning, gene drive, etc; and also a perspective on how humans modified "life" well before the invention of biotech: the dog is a wolf that humans genetically modified over the course of about 30,000 years by selecting the friendliest individuals of the species; in the 18th century the English agriculturist Robert Bakewell invented modern (quasi-scientific) sheep and cattle breeding and created new kinds of cows and sheep; the “gamma ray gardens” of the 1950s (vegetable gardens in which vegetables were exposed to radioactive isotopes) are responsible for 2,700 new varieties of plants, including the peppermint used in today’s mint oil and many orchid, tulip and rose varieties; etc.

Rob Carlson, author of "Biology is Technology" and president of the Bioeconomy Capital

Andrew Endy, Professor of Bioengineering at Stanford University, co-founder of the BioBricks Foundation and of the International Genetically Engineered Machines (iGEM.org) competition

John Cumbers, founder of Synbiobeta and former NASA scientist

Eric Gordon, inventor and venture capitalist of Skyline Ventures

Emily Leproust, founder of Twist Bioscience

Max Hodak, founder of Transcriptic