One of COVID-19’s nastiest tricks is the way it can infect someone and not cause any symptoms. This allows the virus to proliferate under the radar of contact tracers. But new artificial intelligence could help track down these silent carriers.
In a paper published Friday in the journal Scientific Reports, researchers at Synergies Intelligent Systems and Universität Hamburg describe a machine learning algorithm that can identify people in a moving crowd who are most likely asymptomatic carriers of the virus that causes COVID-19. The algorithm makes these predictions based on the GPS-tracked movement of people in a city environment, and known cases of infection.
As new variants of the coronavirus continue to spring up like wildfires across the planet, researchers have been frantically trying to determine which new strains might outwit our brand new vaccines.
Artificial intelligence (AI) may be able to help. In a paper published Friday in the journal Science, researchers at MIT described a machine learning algorithm that can predict which mutations pose the biggest threat to the world’s fledgling immunity.
The tool could be used to quickly narrow down which mutations are most likely to evade the immune systems of people who have been vaccinated or previously infected. Researchers can then test suspected strains in the lab and update vaccines accordingly.
“This is a real-time companion to vaccine development,” says Bryan Bryson, a biological engineer at MIT and co-author of the paper. “What we can do with our model right now is a lot faster than what you can do in the lab.”
As new, more contagious variants of the coronavirus surge across the planet, public health officials are scrambling to increase genetic sequencing of positive samples. Sequencing is crucial in understanding how the virus is changing, and determining whether our brand new vaccines will remain effective, officials say.
“Imagine if we didn’t have this genetic data,” says Richard Neher, a professor at the University of Basel who studies the genetic evolution of viruses. “We would see a surge in cases without having any idea what might have changed.”
While the variants are likely more contagious, there is no evidence suggesting that they are more deadly or cause more severe disease. Many experts also say that the COVID-19 vaccines that have already been developed will still be effective against the new variants. Still, global surveillance of the virus’s genetic sequence is needed to stay on top of the virus’s continual adaptations, and plan a vaccine response.
What are viral variants?
Viruses, including SARS-CoV-2 (the coronavirus that causes COVID-19), are constantly mutating. As they move from person to person, their genetic code changes slightly. Most of these mutations are inconsequential, producing no meaningful changes to the structure or function of the virus.
As a virus moves through populations of people, it begins to accumulate enough mutations to lead researchers to call it a “variant” and give it a name. Many variants of the coronavirus have already been recognized.
Sometimes multiple mutations occur quickly, as was the case with the B.1.1.7 variant in the UK. Neher, who helps track the genetic changes in viruses using the software tool Nextstrain, estimates that B.1.1.7 is about 30-35 mutations away from the original strain detected in Wuhan, China at the beginning of the pandemic. Between 10 and 17 of those mutations appeared suddenly, compared with the virus’s most recent ancestor.
Many of B.1.1.7’s mutations occur in areas of the genome that code for elements of the virus’s spike protein. That’s important, because the virus’s spike protein is what it uses to enter human cells. It’s also what our immune systems will recognize when attacking the virus.
Will approved vaccines work against the new variants?
The more the spike protein changes, the harder it is for the immune systems of people who have been vaccinated to mount a swift attack. The same goes for people who have already had COVID-19—their immune systems know the spike protein of the older variants.
But it takes a lot of genetic changes to the spike protein before it can evade our complex immune systems. “The spike protein is a large protein,” says Neher. “It’s not like a single mutation there would change the virus in a way that it can re-infect everybody on the planet. It’s a more of gradual process where some mutations might reduce efficacy of the immune response in some fraction of the population.”
The section of the genome that codes for the spike protein is about 3,800 nucleotides, or units, long. So even with a dozen mutations, “for all intents and purposes, it’s the same protein,” says Neher.
Many experts, including those at the U.S. Centers for Disease Control and Prevention (CDC) and the U.S. National institute of Allergy and Infectious Diseases, have stated publicly that our current vaccines will most likely be effective against the latest variants.
The mutations are “unlikely to have a large impact on vaccine-induced immunity or on existing immunity” from previous infection, said Greg Armstrong, director of the advanced molecular detection program at the CDC, in a media briefing last week.
If the variants do start to evade immune systems of vaccinated people, vaccines can be altered to mimic the new variants. mRNA-based vaccines, such as those developed by Moderna and Pfizer/BioNTech, can be adjusted relatively quickly.
Public health experts push for more genetic sequencing
But to be sure, the public health community will have to keep a close eye on the variants, as well as the virus’s future adaptions, which will undoubtedly occur. To that end, experts are calling for larger, more coordinated genetic sequencing and epidemiological surveillance.
In a statement posted December 31, the World Health Organization (WHO) advised the world to “increase routine systematic sequencing of SARS-CoV-2 viruses to better understand SARS-CoV-2 transmission and to monitor for the emergence of variants.”
Still, more is needed. On December 29, the CDC said that the U.S. had about 51,000 sequences in its public databases, noting that the UK had more than twice that many. The CDC now aims to scale up to 3,500 whole genome sequences per week, according to Armstrong at the CDC.
To do that, the agency in November launched the National SARS-CoV-2 Strain Surveillance (NS3) program, and asked each U.S. state to send at least ten samples biweekly for sequencing. The agency is also funding and working with national reference labs and local academic centers to increase sequencing.
Armstrong’s division has also been working since 2014 to integrate next-generation sequencing and bioinformatics expertise into state and local health departments. It increased funding for that in December. The effort includes training people to use portable, desktop genetic sequencers such as Oxford Nanopore’s MinION and Illumina’s MiniSeq.
They look like regular earbuds, but these headphones don’t play music, or produce any kind of sound. Instead, they produce electrical fields designed to treat disease.
By delivering electrical pulses to a nerve in the outer ear, the device hacks into neural circuits in the brain in a way that could regulate inflammation and treat rheumatoid arthritis.
That’s the hope, anyway, of researchers at the start-up Nēsos, which launched out of stealth mode today. “We’re still at the early stages of development,” says Konstantinos Alataris, co-founder and CEO of the company. “We’re developing this as a prescription product and testing it in clinical trials.”
And arthritis is just the first application that the startup is pursuing. If Nēsos has found an effective way to hack into the brain, the earbuds could help with a range of neurological and psychiatric diseases.
The competitors were neck-and-neck going into the final turns of the last heat, and in the end, Italy beat Thailand by four seconds. But unlike the Olympic games, none of the competitors in this race could move their bodies. Instead, they competed using only their thoughts.
This is Olympic racing, cyborg-style. Using brain-computer interface (BCI) systems, the competitors—all of whom are paralyzed from the neck down—navigated computer avatars through a racetrack using thought-controlled commands.
The race was part of Cybathlon 2020: The second ever cyborg Olympics, in which people with paralysis or amputated limbs turn themselves into cyborg athletes using robotics and algorithms. Proud competitors raced with their exoskeletons, powered wheelchairs and prosthetic limbs through obstacle courses as their tech teams cheered them on.
The world desperately needs a portable, reliable COVID-19 test that can deliver immediate results. Now scientists at Stanford University say they hope that a new diagnostic assay will fill that void.
In a paper published last week in the journal PNAS, the scientists describe how they used electric fields and the genetic engineering technique CRISPR to build a microfluidic lab-on-a-chip that can detect the novel coronavirus. The test delivers results in about half an hour—a record for CRISPR-based assays, according to the authors—and uses a lower volume of scarce reagents, compared with other CRISPR-based tests in development.
“We’re showing that we have all the elements required to achieve a miniaturized and automated device with no moving parts,” says Juan Santiago, vice chair of mechanical engineering at Stanford, who led the research. He adds, however, that more work lies ahead before their test could be ready for the public.
Two Australian men with neuromuscular disorders regained some personal independence after researchers implanted stent-like electrodes in their brains, allowing them to operate computers using their thoughts.
This is the first time such a device, dubbed a “stentrode,” has been implanted in humans, according to its inventors. The system also makes real-world use of brain-computer interfaces (BCIs)—devices that enable direct communication between the brain and a computer—more feasible.
In his “Plenty of Room at the Bottom” lectureat Caltech in 1959, physicist Richard Feynman urged his audience to make the microscope ever more powerful so that biologists could explore the “staggeringly small world” beyond. It would be a lot easier to answer fundamental biological questions if we could “just look at the thing,” he said.
A few years later, in the science fiction movie Fantastic Voyage, a submarine crew shrinks to microscopic size and goes on a mission through the human body to repair brain damage. The 1966 movie trailer says the film “drops the bottom out of the world you know and understand,” and sends viewers “where no man or camera has gone before.”
Now, scientists have combined the visions of the mid-century physicist and filmmakers in one groovy virtual reality experience. In a paper published last week in Nature Medicine, researchers described new software that enables scientists to enter inside and explore a cell or other biological structures using a virtual reality (VR) headset.
Covid-19 spreads via droplets expelled from an infected person’s lungs, so determining how the release of moisture is affected by different masks is an important step towards better protective gear. Now, using a new technique in 3D printing, University of Cambridge researchers have created tiny, freestanding, conducting fibers they claim can detect respiratory moisture more effectively than anything currently on the market.
The researchers demonstrated the fiber sensors by testing the amount of breath moisture that leaks through face coverings. They attached their fiber array to the outside of the mask, wired it to a computer, and found that it outperformed conventional planer chip-based commercial sensors, particularly when monitoring rapid breathing. (A paper describing the invention was published today in the journal Science Advances.)
Dubbed “inflight fiber printing,” the technique enables the researchers to print the fibers and hook them into a monitoring circuit, all in one step.
“Previously you could have very small conducting fiber production but it could not be incorporated directly into a circuit,” says Shery Huang, a lecturer in bioengineering at the University of Cambridge who led the research. “The main innovation here is we can directly incorporate these small conducting fibers onto the circuit with designable fiber pattern structures,” she says.
In an achievement that would have startled biomedical researchers merely a year ago, vaccines against COVID-19 were already being tested in humans this past March, less than three months after the initial outbreak was identified in China. Many of those vaccines owed their speedy start to the power of artificial intelligence (AI).
The feat is a promising and remarkable turn in the 200-year-plus history of immunization. The experience may revolutionize the way vaccines are created, potentially saving countless lives in epidemics yet to come.
As of early September, there were 34 vaccine candidates being tested in humans, according to the World Health Organization (WHO). Another 145 candidates were being tested in animals or in the lab, says WHO, which keeps a running worldwide list. Those are astonishing numbers, considering that less than a year ago no one had heard of the novel coronavirus, now known as SARS-CoV-2, which causes the respiratory disease COVID-19. It typically takes many years, or even decades, to develop a vaccine; until now, the speed record was held by the mumps vaccine, which went from a collected sample to a marketed product in about four years.
It’s no wonder that research is sprinting ahead. Our societies and economies likely won’t return to normal until a highly effective vaccine has been administered to a substantial portion of the planet’s population. The search for a vaccine is now a vast undertaking, involving thousands of researchers at hundreds of laboratories around the world spending billions of dollars. It’s like a moon shot in its magnitude, ambition, and intensity.
Laboratories are pursuing at least eight different types of vaccine. These include traditional ones based on inactivated viruses, as well as new, more experimental ones involving the use of genetic material—so-called DNA and RNA vaccines—as well as others based on special proteins or other biological agents.
At stake are not only human lives but also a piece of a global vaccine market that was estimated at US $35 billion even before COVID-19. Governments, philanthropies, and pharmaceutical companies have been spending accordingly. In July, the U.S. government agreed to pay pharmaceuticals giant Pfizer and German biotech firm BioNTech nearly $2 billion for 100 million doses of a vaccine, if and when it becomes available. Other major vaccine initiatives worldwide also have funding in the 10 figures.
Machine-learning systems and computational analyses have played an important role in the vaccine quest. These tools are helping researchers understand the virus and its structure, and predict which of its components will provoke an immune response—a key step in vaccine design. They can help scientists choose the elements of potential vaccines and make sense of experimental data. They also help scientists track the virus’s genetic mutations over time, information that will determine any vaccine’s value in the years to come.
“AI is a powerful catalyst,” says Suchi Saria, a professor at the Johns Hopkins Whiting School of Engineering who directs the university’s machine-learning and health care lab. AI enables scientists “to draw insights by combining data from multiple experimental and real-world sources,” she explains. These data sets are often so messy and challenging that scientists historically haven’t even attempted these sorts of analyses, she adds.
As AI tools become more powerful, researchers are anticipating a time when computational methods could help scientists solve our most vexing vaccine challenges—such as finding an effective HIV vaccine, or creating a flu vaccine that’s good for more than a year.
The excitement surrounding new computational techniques comes with a caveat: AI cannot replace or speed up the most crucial, time-consuming aspect of vaccine development. Animal and human trials must happen via pure human effort, with thousands of scientists, health care workers, and participants logging their experience with a vaccine in real time. “Computation helps you optimize your chances of success, but ultimately you have to roll up your sleeves and do it in the lab,” says Jacob Glanville, founding partner at Distributed Bio and its subsidiary, Centivax, which are developing vaccines for flu, HIV, and other pathogens using computational bioengineering.
Still, in the quest for a COVID vaccine, AI has done more than it ever has before. And it is just part of a larger suite of computational tools that are revolutionizing vaccine R&D. Few people may be thinking about the next pandemic, but researchers are already starting to understand how these tools will do quite a bit more the next time around.
Modern vaccinedesign is a hugely information-intensive endeavor, starting with the reams of data needed to understand both the virus and our immune system’s reaction to it. There are more than 200 viruses known to infect human beings, and each of them is distinct in its mechanisms, behavior, and ultimately, its cures.
Though they vary in the details, viral attacks on the body mostly start off the same way. When a virus gets into the body—say, through the mouth or nose—it infiltrates healthy cells by binding to receptors on the cells’ surfaces. The virus can then hijack the cells’ machinery to make more copies of itself, and an infection ensues.
Putting a halt to all this is the job of the immune system, which hunts and destroys pathogens such as viruses and bacteria that cause disease. As a first step, the immune system sends a variety of basic weapons to the infection in what’s called the innate immune response.
If that’s not enough to control the infection—for example, if the pathogen is new to our bodies and the generalized weapons don’t work—the immune system’s adaptive immune response brings in the bigger guns. Adaptive immunity depends on two types of white blood cells, called B cells and T cells. B cells produce specialized proteins called antibodies, which bind to the pathogen and prevent it from entering healthy cells. T cells, meanwhile, can destroy cells that have been infected by the virus, to keep them from making more copies of it.
It takes days for the adaptive immune response to get revved up enough to begin wiping out a new virus. Our bodies have B cells and T cells tailored for nearly every pathogen the world can throw at us, but it takes time for the right immune cells to find the invader and multiply. In the meantime, we get sick.
The good news is that while this war is raging, the immune system also produces memory B and T cells, which make a record of the battles. If we get exposed to the same pathogen again, the immune system has an arsenal at the ready and responds much more rapidly. We may experience mild symptoms, or none at all.
The goal of a vaccine, then, is to expose the body to a pathogen without making us sick so that the immune system is primed to fight it on any subsequent exposure. This can be done by exposing the body to specific pieces of a virus or weakened versions of it. Crucially, the vaccine must include key parts of the virus, called antigens, that are immunogenic, meaning that they’re recognizable to B cells and T cells and will therefore trigger the desired adaptive immune response.
When faced with a new pathogen, the first question for vaccine designers is: Which parts of it are the most immunogenic? A typical virus consists of genetic material, either DNA or RNA, encapsulated by one or more layers of proteins. The outer membrane is often studded with so-called spike proteins, which enable the virus to bind to receptors on a host cell and inject its payload of genetic material. For this reason, spike proteins are a typical target for vaccines. If the immune system creates antibodies that disable the spike protein, the virus cannot break into cells.
However, for any given virus there are tens of thousands of different subcomponents of the outer proteins that the immune system can recognize, and therefore tens of thousands of different possibilities for vaccine targeting. This is a prime opportunity for AI. Machine-learning tools can predict, based on training data sets from known pathogens, which pieces of the virus the immune system is most likely to recognize.
Armed with this information, immunologists can design vaccines around a more manageable number of potential targets. The targets are then integrated into vaccine candidates and tested in animals to see if they provoke a good immune response. Machine learning “gives you a numerical score,” says Tayab Waseem, a public policy fellow at the American Association of Immunologists and director of medical informatics and AI integration at Wagner Macula & Retina Center, based in Norfolk, Va. “Everything over a certain score—say, 99 percent—I’ll be willing to go into the lab to test out.”
As expected, many of the system’s top recommended targets were located on the virus’s spike protein. Chen’s team recommended, in a paper on the preprint server bioRxiv, that these epitopes be included in the design of COVID-19 vaccines. “We feel pretty confident that we’ll get an immune response at the cellular level against what we predicted as targets,” says Chen. “But there is a big gap between cellular response and clinical response,” he adds.
Chen’s machine-learning tools are among several dozen that have been built over the years to aid immunology work. In the past, machine learning has been a “minor sidekick” in vaccine development, according to Chen. But for COVID-19, “people in both academic and industry labs ran more computational studies,” he says. “I suspect that all the pharma companies who developed a vaccine also ran a computational analysis.”
Having identified a target on the virus’s surface, researchers can then develop a vaccine. If the plan is to use an inactivated virus as a vaccine, for example, researchers will grow the live virus in the lab and kill it using heat, radiation, or a chemical method so that it can’t replicate when injected into the body. Then researchers must make sure that the key immunogenic components weren’t damaged when the virus was killed, as those parts must be intact in order to provoke an immune response. The next steps are to test the vaccine in the lab, then in small animals, and finally in humans.
To train software to sift through target sites on a virus, it’s important to first understand the three-dimensional structure of viral proteins. Viral proteins are made of linear chains of chemicals called amino acids, which spontaneously fold into compact, ribbonlike structures. Vaccine developers must choose targets on the virus’s outer layer that face outward, so that they’re physically accessible to immune-system weaponry.
When the pandemic hit, researchers at the University of Basel, in Switzerland, used a protein-modeling tool called Swiss-Model to predict the structures of the proteins on the outer surface of the SARS-CoV-2 virus. Their predictions were later shown to be consistent with the virus’s actual protein structures. Similarly, the London-based AI company DeepMind applied its neural network, AlphaFold, to predict the three-dimensional shape of SARS-CoV-2 proteins based on the virus’s genetic sequence.
Despite these successes, not all researchers are enthusiastic about the promise of AI for this component of vaccine R&D. They note that, AI or no AI, the spike protein was an obvious target, based on knowledge of other coronaviruses and experimental work with SARS-CoV-2. “There are a lot of methods to identify immunogenic regions of pathogens that do not require artificial intelligence,” says Glanville of Distributed Bio. Algorithms that predict such targets are nice to have, he says, but probably not necessary in the case of COVID-19. “AI still has the challenge of proving that it works better than simpler methods,” such as serological screening, epitope mapping, and structural biology, he says.
But AI can do much more than zero in on the immunogenic sites on a virus. Many vaccine developers are already using computational tools to design and synthesize the genetic components of DNA-based vaccines. Inovio Pharmaceuticals in San Diego, one of the 34 groups with a COVID-19 vaccine in human trials, is one example.
“The team at Inovio waited enthusiastically for the genetic sequence of the virus to be posted online,” says Kate Broderick, senior vice president of R&D at Inovio. “When it was uploaded by the Chinese authorities on January 10, our scientists immediately entered the sequence into our algorithm, and within 3 hours they had a fully designed and optimized DNA medicine vaccine,” she says.
Inovio’s DNA vaccines work by mimicking a part of the genetic sequence of the pathogen. These so-called nucleic-acid vaccines contain segments of genetic instructions, in the form of DNA or RNA, that code for a key immunogenic component of the virus. When the nucleic acid is inserted into human cells, the cells produce the antigen, which triggers an immune response. Inovio researchers knew, based on previous research on other coronaviruses, that the spike protein of SARS-CoV-2 would likely elicit an immune response. So that region of the virus’s genome became their starting point for a vaccine.
There are many different ways to write out a DNA sequence that codes for the production of the same protein. To find the one that will work best as a vaccine, that bit of code has to be enhanced with other genetic and molecular elements. Inovio’s proprietary gene-optimization algorithm showed researchers how to do this in such a way that the vaccine would provoke the large-scale production of an immunogenic spike protein.
Inovio’s COVID-19 vaccine went from bench to bedside in just 83 days, Broderick says. The vaccine performed well in animals—as shown by a study that she and her colleagues published in May in Nature Communications. In late June, the company announced that the vaccine proved safe and appeared to provoke immune responses in 40 healthy people in a trial in the United States. The vaccine also provided protection for four months in monkeys that were vaccinated and then exposed to the virus, according to a report from Inovio in late July.
Keeping up with a virus’s genetic changes also presents a challenge well-suited for computational analysis. Viruses are constantly mutating in small ways, so a vaccine must be designed around a relatively stable region of the virus’s genome—a region of its genetic code that doesn’t tend to mutate. “There are certain parts of the surface proteins on the virus that have a very high turnover, which you only find out as you sequence it and get changes in structure as it mutates,” says Waseem of the Wagner Macula & Retina Center.
Over the last 10 months, tens of thousands of COVID-19 virus samples taken from patients around the world have been genetically sequenced and uploaded into an online repository hosted by the Global Initiative on Sharing All Influenza Data (GISAID), in Germany. Algorithms that compare those sequences can reveal which segments of the virus’s genome frequently change, and which segments don’t. As the virus continues to conquer new territories, researchers will keep tabs on their ever-changing foe.
All of this work takes a lot of computing power. In March, the White House announced that it would collaborate with public and private groups to provide researchers worldwide with access to the most powerful supercomputers in an effort to “rapidly advance scientific research for treatments and a vaccine.”
Mahmoud Moradi, a computational chemist at the University of Arkansas, in Fayetteville, led one of those projects. He used supercomputers at the Texas Advanced Computing Center to create enhanced 3D simulations of coronavirus spike proteins. The simulations revealed that the spike proteins become active and infect human cells much faster than those of a previous infectious coronavirus, SARS-CoV-1, which caused an outbreak in Asia in 2003.
Broderick at Inovio says that this kind of research is vital to vaccine-development teams. “Scientists can learn a huge amount of relevant information to assist with vaccine design,” she says, “as well as understanding the mechanisms behind the pathogenesis of this virus.”
Once a vaccine candidate is designed, the bulk of the work then shifts to testing. Vaccines are first tested in the lab on cells and on animals, and then on increasing numbers of people in clinical tests. Tens of thousands of trial volunteers will have received a vaccine before it’s approved by U.S. regulators.
Unfortunately, AI tools can’t replace those time-consuming steps. They might be able to predict which antigens the immune system will see, but what the immune system will actually do, in a live human, is beyond the capabilities of today’s computers. “The human body is so complex that our models cannot necessarily predict with reliability what this molecule or this vaccine will do for the body,” says Oren Etzioni, CEO at the Allen Institute for Artificial Intelligence. “That’s why we have these slow and painful trials—our predictive models aren’t good enough to give you reliable data.”
Although AI can’t predict the success of human trials, it can make sense of the mountains of data from these experiments by looking at all the parameters and finding patterns that a human brain might not spot. As vaccine candidates advance to second and third phases of clinical testing, thousands of patients will be involved, and AI systems will be key in rapidly analyzing the clinical and immunological data.
And as more researchers add their studies to the ever-increasing body of literature on the novel coronavirus, scientists will need help sorting through those papers. The Allen Institute developed a resource called CORD-19 that provides more than 130,000 scholarly articles on COVID-19 in machine-readable format. The Kaggle community, among other groups, leveraged the data set to create multiple AI systems to help researchers keep up with literature and answer high-priority research questions.
“I believe that within a decade AI will be an indispensable part of any medical researcher’s tool kit both for scouring the literature and for analyzing experimental data,” says Etzioni. And when the next pandemic comes—because there will always be a next pandemic—researchers will be poised to unlock the secrets of the deadly pathogen, design many potential vaccines to protect us, and rapidly identify the ones that can prevent a disaster like COVID-19 from befalling humanity again.
This article appears in the October 2020 print issue as “AI Takes Its Best Shot.”
Researchers are focusing more and more attention on using artificial intelligence to discover and design new medicines, and a new start-up with strong backing announced last week that it is jumping into the field with big ambitions.
Cambridge, Massachusetts–based Generate Biomedicines, which emerged from stealth mode on 10 September, says it is using machine learning algorithms and big data to design biological compounds to combat 50 of the top targets associated with disease. The company is also developing candidates for therapies that would fight SARS-CoV-2, the virus that causes COVID-19.
Generate Biomedicines is backed by Flagship Pioneering, a funder and builder of start-up companies. It’s roster includes Moderna, which is a leader in the race to develop a vaccine against COVID-19.
“We are creating new molecular entities, and generating molecules that you wouldn’t be able to discover through traditional means,” says Molly Gibson, co-founder and chief innovation officer at Generate Biomedicines. “We think this type of technology breaks you away from that discovery paradigm of searching through existing molecules or searching through close relatives of existing molecules.”
Traditional protein drug discovery methods, such as high-throughput screening, involve a lot of trial and error. Algorithms and high powered computing could greatly shrink the amount of time scientists spend searching for new drug candidates.
“If you can compute sequences in silico, you’re speeding up the time it takes to come up with an effective candidate,” says Gini Deshpande, founder and CEO of AI-based drug discovery company NuMedii, who is not involved with Generate Biomedicines. “There are a number of players in this space coming at it from different angles,” Deshpande says, noting that there is a huge unmet need for technologies that can reduce the time and cost of drug development.
Deshpande says that there are at least 17 different companies using machine learning to aid the discovery of biologics, or protein-based drugs—the kind of medicines on which Generate Biomedicines is focused. Deshpande estimates that over 200 more companies are applying AI to small molecules—a category of drugs characterized by their low molecular weight. “So they’re certainly not the only player in the space,” she says.
Avak Kahvejian, founding co-CEO of Generate Biomedicines, says his company stands out in part because its machine learning platform is capable of teasing out “the true, underlying foundational principles by which proteins operate,” he says. Other companies “are using machine learning to find useful relationships in data.”
The company’s computational platform is trained on a vast database of known proteins: 160,000 protein structures and 190 million protein genetic sequences, according the company. The system looks for statistical patterns linking a protein’s genetic sequence, three-dimensional structure, and function, and develops a set of governing rules for those patterns—similar to the way algorithms process natural language and images.
After learning, the system can then be focused on the synthesis of new, custom proteins with therapeutic potential. The platform continually learns from every new protein generation campaign through a generate-build-test cycle.
“With this vast amount of training and learning, we can now run the process forward, and use it to generate a number of beautiful solutions that are unimaginable today—undesignable by the human hand,” says Kahvejian. “These general rules are much more valuable, as it means you don’t have to reinvent the wheel every time you see a new class of proteins. It also means you can immediately sample from all the sets of protein classes that nature never invented.”
Generate Biomedicines’s drug candidates will include proteins of any variety—such as antibodies, peptides, enzymes, and cytokines—that can bind to, disable, or activate biological targets associated with disease. The company aims to take its drugs all the way through human trials and to market, says Kahvejian. “The breadth [of possible therapies] is enormous and daunting, and it’s nothing one company can really commercialize in and of itself,” he says. “So partnering will be part of the strategy.”
In response to the COVID-19 pandemic, Generate Biomedicines has applied its platform to designing antibodies and peptides that can neutralize the SARS-CoV-2 virus. “We immediately jumped on trying to figure out what’s known about the virus and the protein structures involved in its infectivity, and use our platform to generate antibodies to block those structures,” says Kahvejian. Those antibodies are now being tested in the lab in collaboration with the Coronavirus Immunotherapy Consortium, he says.
Celebrity engineer Elon Musk today announced a breakthrough in his endeavor to sync the human brain with artificial intelligence. During a live-streamed demonstration involving farm animals and a stage, Musk said that his company Neuralink had built a self-contained neural implant that can wirelessly transmit detailed brain activity without the aid of external hardware.
Musk demonstrated the device with live pigs, one of which had the implant in its brain. A screen above the pig streamed the electrical brain activity being registered by the device. “It’s like a Fitbit in your skull with tiny wires,” Musk said in his presentation. “You need an electrical thing to solve an electrical problem.”
Musk’s goal is to build a neural implant that can sync up the human brain with AI, enabling humans to control computers, prosthetic limbs, and other machines using only thoughts. When asked during the live Q&A whether the device would ever be used for gaming, Musk answered an emphatic “yes.”
Musk’s aspirations for this brain-computer interface (BCI) system are to be able to read and write from millions of neurons in the brain, translating human thought into computer commands, and vice versa. And it would all happen on a small, wireless, battery-powered implant unseen from the outside of the body. His company has been working on the technology for about four years.
Teams of researchers globally have been experimenting with surgically implanted BCI systems in humans for over 15 years. The BrainGate consortium and other groups have used BCI to enable people with neurologic diseases and paralysis to operate tablets, type eight words per minute and control prosthetic limbs using only their thoughts.
All of this work is highly experimental. Since 2003, fewer than 20 people in the U.S. have received a BCI implant, all for restorative, medical purposes on a research basis. Most of these systems involve hardware protruding from the head, providing power and data transmission.
These external components create the potential risk of infection and aren’t practical outside a research setting. A few groups have experimented in animals with self-contained, fully implanted devices, but not with the capabilities that Neuralink claims to have.
Neuralink’s implant contains all the necessary components, including a battery, processing chip, and bluetooth radio, along with about a thousand electrode contacts, all on board the device. Each electrode records the activity of somewhere between zero and four neurons in the brain. A thousand of them in a living animal would be the highest number the BCI field has seen from a self-contained implant.
Neuralink’s device, if it proves capable of transmitting data safely over the long-term, would be a “major advance” says Bolu Ajiboye, an associate professor of biomedical engineering at Case Western Reserve University and a principal investigator with BrainGate, who is not involved with Neuralink. “There are some really smart, innovative people working at Neuralink. They know what they’re doing and I’m excited to see what they present,” he says.
But the company’s data has not yet been vetted by the research community. (Three pigs on a stage isn’t quite the same as peer-reviewed data). How the device can transmit that much data without generating tissue-damaging heat is not yet demonstrated in humans.
Plus, Neuralink’s device is “pretty big” for the brain, says Ajiboye. Its cylindrical shape measures 23 mm in diameter by 8 mm long—about the size of a stack of 5 U.S. quarters. By comparison, the Utah array, which has been the go-to device for the BrainGate consortium, measures 4 mm x 4 mm. That device involves hardware protruding from the skull and contains about a hundred electrodes, compared to Neuralink’s 1000.
Neuralink achieved the advance by experimenting with different materials, upgrading the antennae and wirelessly transmitting only heavily compressed embeddings of neural data from the implant, along with other optimizations made possible through a fast feedback cycle, says Max Hodak, president of Neuralink, who spoke with Spectrum prior to today’s live demonstration. One of the company’s latest prototypes is made of monolithically cast forms of glass that are laser welded together and hermetically sealed. The device so far has lasted safely in pigs for two months, says Hodak.
During today’s demonstration, which was held at Neuralink’s headquarters in Fremont, California, three pigs were led into corrals where they were able to move about freely in front of a small (human) audience. Gertrude, the pig with the implant, didn’t want to come out to her corral at first, leaving Musk stranded in front of over 150,000 online viewers. She did eventually come out, with her brain activity streamed on a screen above her. Every time she sniffed the electrical activity in her brain spiked.
Once this kind of brain wave data is obtained, the big question is how to decode and interpret it. “Neural decoding is critically important,” says Ajiboye. “A number of laboratories around the world are spending lots of person-hours on decoding algorithms, using different statistical and deep learning approaches. I haven’t seen that from Neuralink.”
Neuralink has developed a surgical robot capable of inserting the implant’s electrodes at shallow depths into the brain. Robotic precision reduces the risk of damage to brain tissue.
Neuralink’s first applications for the technology will be for medical purposes, likely for people with spinal cord injuries. Musk, in bold fashion, has said he wants to pursue non-medical applications too, further in the future. This has led to a lot of hype in the media.
“We as a field need to be very responsible about what we’re claiming the technology can do, and what application we’re driving toward,” says Ajiboye. “By Elon Musk being in this field there’s a lot of attention being brought to it. That is welcome, but there are challenges posed there. One of those challenges is hype versus reality.” He adds: “Neuralink has entered this race and is riding a fast horse, but there are other devices in development.”
After nearly six months of scrutinizing the smorgasbord of COVID-19 tests available globally, an independent diagnostics evaluation group has come to some conclusions: The world greatly needs rapid, affordable tests, but the quality of such tests varies widely. By contrast, slow tests that are sent off to laboratories—the kind broadly used in the U.S. and Europe—are performing well.
That’s according to the Foundation for Innovative New Diagnostics, or FIND, headquartered in Geneva, Switzerland. “People in low- and middle-income countries often do not have access to laboratories,” says Jilian Sacks, FIND’s COVID-19 Evaluation Programme Lead. “So the ability to have rapid tests, especially outside the hospital and in decentralized settings, is most critical moving forward. But we are seeing that there is a lot of variability in the performance of rapid tests,” she says.
Sacks’s organization in February took on the ambitious task of evaluating the performance of hundreds of COVID-19 tests that have been flooding the global market ever since the pandemic hit. FIND told Spectrum earlier this year that it would subject these tests to a slew of coronavirus samples and make the results public on its website. This week, we checked in to see how it’s going.
The most reliable tests, FIND has found, are molecular tests performed in a laboratory. These diagnostics are the go-to COVID-19 testing method in the U.S. and Europe. They identify the disease by looking for the virus’s genetic code using a tool called polymerase chain reaction, or PCR.
FIND has evaluated 23 molecular tests so far. All but two had a sensitivity of at least 95%, meaning the tests accurately detected positive samples 95% of the time. “We’re seeing that generally, the companies that we selected to evaluate have quite good performance with their assays,” so there are several good options from which ministries of health can choose, Sacks says.
But molecular tests are slow. PCR takes only a few hours to perform, but people are having to wait days to get results back due to the logistics of transporting samples to a centralized laboratory with PCR capability.
“Global demand has continued to soar, and though there may be many companies that should be able to meet that demand, it will require continued, unprecedented scale-up of their manufacturing and distribution,” Sacks says.
Rapid tests, on the other hand, provide results right away, on the spot, without having to send off a sample to a centralized lab. This allows people to make decisions immediately about whether or not to quarantine. And rapid tests are sometimes the only way to get tested in low-resource regions where access to centralized laboratories is scarce. But these tests are not as accurate as molecular tests.
FIND is evaluating 41 different rapid tests—5 that spot antigens and 36 that look for antibodies. Antigens are the parts of the virus that the immune system recognizes, and finding evidence of them means the virus is present. Antibodies are produced from healthy cells as part of the body’s immune response to a virus, and finding those means the virus is, or was, in the body, and that the immune system responded. Both types of tests can be performed on-the-spot because they are less complex and don’t require a lab.
FIND has released results for four of the rapid antibody tests it is evaluating. Only one of them—an antibody test made by BTNX—correctly identified positive cases over 90% of the time. The performance of some of the other tests is so low that they provide little value.
In addition to its own program, FIND is aggregating the independent evaluations of other organizations from around the world and making it available online. These data are sorely needed. Global regulators, such as the FDA, which would normally require a lot more data from diagnostic companies before allowing a test to be used commercially, have largely stepped out of the way. This has boosted innovation and hastened the availability of tests. But it also allows inaccurate tests onto the market, and shifts the burden of oversight to whomever wants to take on the job.
That’s the hole that FIND is trying to fill. “We have indeed heard from various ministries of health, that as companies are approaching them to offer their tests, they are looking to our website to see if we’ve evaluated the test and whether they feel comfortable with the results,” Sacks says.
It’s good to know that someone out there is testing the tests.
For more than three centuries scientists have believed that human sperm swim by swishing their tails in a side-to-side, symmetrical motion. But that’s because we’ve been looking at them with 2D microscopes.
Using state-of-the-art 3D microscopy, a piezoelectric device, and mathematics, researchers in Mexico discovered how sperm really move: They spin, with a wonky asymmetrical wiggle. The researchers reported their discovery today in the journal Science Advances.
An astronaut aboard the International Space Station (ISS) has successfully assembled human cartilage using the power of magnetism.
The feat was achieved using a magnetic levitation bioassembly device installed onboard the station. The machine enables clusters of human cells to assemble into tissue structures, without the use of a physical scaffold. The experiment was described in a paper published today in the journal Science Advances.
“One could imagine not too far in the future that if we colonize Mars or do long-term space travel, we might want to do experiments where we build functional tissues in space, and test them in extraterrestrial environments,” says Utkan Demirci, a researcher at Stanford and an author of the paper.
Imagine getting a space-related injury that rips off your skin or bone, and being able to patch it up with bioengineered tissue—like the movie “Ad Astra,” where people live, work and receive medical treatment on the red planet.
The temperature of your body is no longer considered private information. That’s the stance that businesses around the world are taking as they install thermal imaging cameras, often equipped with facial recognition technology, in their buildings in an attempt to cope with the COVID-19 pandemic.
Airports, office buildings, fast food restaurants, government offices, hospitals, shopping centers, universities—all sorts of places are rapidly adopting the technology. It’s a movement one might call the automation of temperature checks.
The goal is to spot—and turn away—anyone walking into an establishment with a fever. This theoretically reduces the spread of the virus and brings some peace of mind to people in the building. Or, at the very least, it provides some legal protection for the establishments, which can point to the technologies to show that they’ve taken measures to protect their occupants from the virus.
As businesses scramble to find ways to make workers and customers feel safe about entering enclosed spaces during a pandemic, several companies have proposed a solution: COVID-19 air monitoring devices.
These devices suck in large quantities of air and trap aerosolized virus particles and anything else that’s present. The contents are then tested for the presence of the novel coronavirus, also known as SARS-CoV-2, which causes COVID-19.
Several companies in the air quality and diagnostics sectors have quickly developed this sort of tech, with various iterations available on the market. The devices can be used anywhere, such as office buildings, airplanes, hospitals, schools and nursing homes, these companies say.
But the devices don’t deliver results in real time—they don’t beep to alert people nearby that the virus has been detected. Instead, the collected samples must be sent to a lab to be analyzed, typically with a method called PCR, or polymerase chain reaction.
This process takes hours. Add to that the logistics of physically transporting the samples to a lab and it could be a day or more before results are available. Still, there’s value in day-old air quality information, say developers of this type of technology.
“It’s not solving everything about COVID-19,” says Milan Patel, CEO of PathogenDx, a DNA-based testing company. But it does enable businesses to spot the presence of the virus without relying on people to self-report, and brings peace of mind to everyone involved, he says. “If you’re going into a building, wouldn’t it be great to know that they’re doing environmental monitoring?” Patel says.
Patel says he envisions the device proving particularly useful on airplanes, in large office buildings and health care facilities. On an airplane, for example, if the device picks up the presence of the virus during a flight, the airline can let passengers on that plane know that they were potentially exposed, he says. Or if the test comes back negative for the flight, the airline can “know that they didn’t just infect 267 passengers,” says Patel.
In large office buildings, daily air sampling can give building managers a tool for early detection of the virus. As soon as the tests start coming back positive, the office managers could ask employees to work from home for a couple of weeks. Hospitals could use the device to track trends, identify trouble spots, and alert patients and staff of exposures.
Considering that many carriers of the virus don’t know they have it, or may be reluctant to report positive test results to every business they’ve visited, air monitoring could alert people to potential exposures in a way that contact tracing can’t.
Other companies globally are putting forth their iterations on SARS-CoV-2 air monitoring. Sartorius in Göttingen, Germany says its device was used to analyze the air in two hospitals in Wuhan, China. (Results: The concentration of the virus in isolation wards and ventilated patient rooms was very low, but was higher in the toilet areas used by the patients.)
Assured Bio Labs in Oak Ridge, Tennessee markets its air monitoring device as a way to help the American workforce get back to business. InnovaPrep in Missouri offers an air sampling kit called the Bobcat, and Eurofins Scientific in Luxembourg touts labs worldwide that can analyze such samples.
But none of the commercially available tests can offer real-time results. That’s something that Jing Wang and Guangyu Qiu at the Swiss Federal Institute of Technology (ETH Zurich) and Swiss Federal Laboratories for Materials Science and Technology, or Empa, are working on.
They’ve come up with a plasmonic photothermal biosensor that can detect the presence of SARS-CoV-2 without the need for PCR. Qiu, a sensor engineer and postdoc at ETH Zurich and Empa, says that with some more work, the device could provide results within 15 minutes to an hour. “We’re trying to simplify it to a lab on a chip,” says Qiu.
As thousands of individuals continue to gather in U.S. cities to protest racial injustice, computer scientists whose models predict the spread of COVID-19 are considering how these mass gatherings might affect the rate of disease transmission.
As shuttered businesses make plans to resume on-site operations, many plan to outfit their employees with new, anti-pandemic gear: wearable tech that could prevent the spread of COVID-19 inside the workplace.
Other employers are considering equipping their workforces with wearables—separate from their phones—that are capable of granular on-site and indoor location tracking and contact tracing. CarePredict recently rolled out such devices for senior living facilities.
In fact, in a survey of 871 finance executives at companies in 24 countries, 21 percent said they were eyeing location tracking and contact tracing for their workforces, according to PwC, which conducted the survey and posted it online this week.
How we move about in our communities—where we go and how often—greatly affects the spread of COVID-19. And few know our whereabouts better than Facebook and Google.
So, in an effort to help researchers combat the pandemic, the two companies say they are now making their troves of GPS-based mobility data available. The data comes from users who opt in to location services on the companies’ platforms and is provided for public health use in an aggregated, anonymized way.
Such data is vital to public health researchers’ efforts to understand trends in population movement and predict the spread of the disease, which is caused by the novel coronavirus SARS-CoV-2. Local government officials can use the data to make informed decisions on travel and social distancing interventions.
Data for Good
Both Facebook and Google are providing information about where people are going, but the companies differ in the way they are releasing the information.
Facebook, through its Data for Good program, provides mobility datasets and maps directly to researchers upon request. Facebook generates the data in file formats that support epidemiological models and case data.
“We’re sharing the data in a way that public health researchers can use,” says Laura McGorman, policy lead for Facebook’s Data for Good program. “Once a researcher signs a license agreement, they can request data through our mapping portal and get it the next day,” she says.
The mobility datasets let researchers look at population movement between two points, movement patterns such as commutes, and whether people are staying close to home or visiting many parts of town. Facebook’s is the“only source of mobility data in machine-readable format” that is global and free of charge, says McGorman.
Data for Good started three years ago as an initiative to help track evacuations and displacement after natural disasters. It has since expanded to address disease and, most recently, COVID-19. The company gathers its information from people using Facebook on their mobile phones with the location history feature enabled. Data is aggregated to protect individual privacy.
Scientists have used Facebook’s data in several ways over the last few weeks to study the pandemic. For example, scientists at the Institute for Disease Modeling in Bellevue, Washington used Facebook’s mobility data to study how social distancing measures and a stay-at-home orders have affected movement near Seattle. They found that population movement indeed declined, which led to reduced transmission of the virus.
Separately, researchers in Italy used Facebook’s mobility data to analyze how lockdown orders affect economic conditions and create an economic segregation effect. A report from the National Tsing Hua University in Taiwan used Facebook’s data to show that travel restrictions reduced the spread of the virus.
Facebook’s program is also supplying the bulk of the data for theCOVID-19 Mobility Data Network. The recently-formed group, composed of a network of epidemiologists, uses mobility data to generate daily situation reports for decision makers who are implementing social distancing interventions.
Google’s mobility tracking tool
Separately, Google on April 3 announced that it had launched a mobility tracking tool called COVID-19 Community Mobility Reports. The web-based tool is available freely to the public and provides insights on how communities have reduced or increased their visits to certain types of places.
The public can go to the website and choose a region, such as a state or country. The tool then generates graphs on a downloadable PDF displaying the percentage change in visits over the last few weeks to places such as retail stories, pharmacies, parks, places of work and public transportation hubs in that region.
In the county where Indianapolis is located, for example, people have reduced their visits to grocery and pharmacy stories by 17% and to other retail locations by 45%, since February 23. Visits to parks, however, have increased 54%.
In a blog post highlighting the resource, Google executives wrote that they believe the mobility reports could help shape business hours, inform delivery service offerings, or indicate a need to add additional buses or trains to a particular public transportation hub.
The company pulls the data from Google users who have opted in to location tracking services. The information is aggregated and anonymized, and does not provide real-time data in an effort to protect privacy.
Mobility data similar to that from Facebook and Google have already informed decisions of government officials. Tennessee Governor Bill Lee on April 2 issued a statewide order for residents to stay at home after he reviewed mobility data released by tech startup Unacast. The information, gleaned from mobile phone location data, showed that people in some regions, such as Nashville, had significantly reduced their daily travel, but people in many other Tennessee counties had not. This convinced Lee that a statewide order was necessary.
Both Facebook and Google are releasing other kinds of data to coronavirus researchers and the public. Data for Good offers population density maps as well as social connectedness indices. The latter relies on aggregated, anonymous friendship connections on Facebook to measure the general connectedness of two geographic regions.
That type of information can help predict the spread of the virus and where to put resources. Researchers at NYU used the social connectedness data to show that geographic regions with strong social ties to two early COVID-19 hotspots—in New York and Italy—had higher cases of the illness. Separately, an organization funded by the World Bank used Facebook’s population density data to help determine where coronavirus testing facilities and extra beds should be located.
Google and Apple last week announced an ambitious effort to provide the technological support for digital contact tracing. The strategy would allow people with certain Bluetooth-enabled apps to find out if they have been in the vicinity of people who have tested positive for the novel coronavirus.
Digital contact tracing has been touted by public health specialists as a strategy to help reopen the economy in a safe way, but privacy and ethical considerations have been hotly debated.
The collective thoughts of the interwebz
The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.