1910 Genetics
In 1910, a decade after Gregor Mendel’s pioneering discoveries in genetics were rediscovered, James Bryan Herrick, MD, published the first case report of a patient with sickle cell anemia in Western medical literature. Walter Clement Noel was a Grenada-born dental student treated for respiratory problems at Chicago’s Presbyterian Hospital by Herrick’s intern, Ernest Irons, who noticed that a smear of Noel’s blood contained “many pear-shaped and elongated forms—some small.” Noel died of asthenic pneumonia in 1916 at age 32.
Almost a century later, Herrick’s case study impressed Jen Nwankwo, PhD, who grew up in Nigeria before emigrating to the United States to pursue her education as a pharmacologist, which included studying sickle-cell anemia, among other diseases. Her PhD research led to the discovery of a novel drug target for sickle-cell disease (SCD). Later, she used this target to validate the technology of the AI-based drug developer she officially incorporated in 2018—and named for the year of Herrick’s famous discovery.
Based in Cambridge, MA, 1910 Genetics recently emerged from stealth mode by completing a $22 million Series A financing led by Microsoft’s venture capital fund M12 and Silicon Valley early-stage investment firm Playground Global.
Proceeds from the financing will help scale up 1910’s AI and biological automation platforms, building its team, expanding partnerships with biopharma companies, and advancing its pipeline of therapeutic programs.
Nwanko told GEN Edge that she was and remains inspired by Herrick’s story. “When 1910 Genetics goes after targets, I want us to have that level of clarity,” she said. “We’re not a sickle cell company. Sickle cell was a disease that I grew up around, and it’s a common killer of children in Nigeria and in most of suburban Africa. It’s the first disease for which we have a clear molecular basis, the first time that as scientists we could say, ‘Here’s this phenomenon, this pathology in a human being.’”
Between 20% and 30% of Nigerian adults carry the SCD trait, while Nigeria has the world’s highest numbers of people with the disease (643,374 disability-adjusted life years per 100,000) and people who die from its complications (more than 7,100 per 100,000), according to the Institute for Health Metrics and Evaluation at the University of Washington.
In the United States, Nwankwo worked as an R&D associate with Eli Lilly and Novartis, then headed business development at Transparency Life Sciences before completing her PhD at Tufts University School of Medicine in 2016. She later advised biopharmas as a management consultant with Bain Capital in Boston, where she developed an interest in AI deep enough to want to launch her own business based on the technology.
Pain points
“The idea for 1910 Genetics came out of what I saw as an opportunity for technology to help address pain points that I was experiencing as a pharmacologist,” Nwankwo said.
Those pain points included the trial-and-error nature of traditional drug discovery; she once screened a library of two million compounds, only to find a solitary potential target.
“I quickly saw that AI—and machine learning as a subset of that—could really help us take advantage of our historical failures. Machine learning could help us learn from those failures in a more intelligent way. It actually contextualized our successes better,” Nwankwo said. “Often, even when we’re successful, it’s almost as if we’re successful in spite of ourselves.”
Nwankwo led 1910 Genetics toward acceptance into seed-funding startup accelerator Y Combinator, followed by acceptance into LabCentral life science incubator in Cambridge. 1910 also won a previously unannounced $4.1 million seed round led by a personal investment from Y Combinator’s former president, Sam Altman, CEO of OpenAI, with participation from Y Combinator, FoundersX Ventures, Scientia Ventures, Emles Advisors, Tuck Lye Koh of Shunwei Capital, and other investors.
Nwankwo was not looking to raise Series A funding but pursued it after connecting with M12 through a longtime friend , Kouki Harasaki, PhD, a former partner at Romulus Capital and later Andreessen Horowitz, who joined Microsoft last year to oversee M12’s healthcare and life sciences investing.
Harasaki met Nwankwo in 2012 when he addressed a biotech club she headed at Tufts during her PhD, to discuss how students could apply their PhDs beyond being scientists. He was an R&D scientist at Novartis at the time, transitioning into a business development position.
“We connected over a shared sense of duty to apply our respective PhDs to help patients,” Harasaki said on M12’s blog announcing the Series A financing. “Since then, she has won my deep admiration for staying true to her passions and for her strong drive.”
Challenge from Microsoft
Microsoft had given Harasaki the challenge of mapping out how to invest in computational drug discovery. “He wanted to do computational drug discovery and he wanted to invest in a founder that he considered strong and he knew,” Nwankwo said. She met both criteria.
M12 officially reached out to 1910 Genetics in September, offering a formal term sheet about eight weeks later. Playground joined in December 2020, and then the round officially closed in February 2021.
Before joining M12, Harasaki had worked on financings of several AI-focused drug developers at Andreessen after it expanded into life sciences in 2015 with its first $200 million Bio Fund—including investments in Insitro and twoXAR.
“[Harasaki] wanted to understand what I was up to and how 1910 was going to differentiate in what is an increasingly crowded space,” Nwankwo said.
One way is by keeping its research focused and in-house. Nwankwo contrasted 1910 Genetics’ approach with that of San Francisco AI-based drug developer Atomwise, which is building a portfolio of joint ventures with startups and researchers based on research programs spanning oncology, immunology, infectious disease, neuroscience, and clotting disorders.
Atomwise’s programs apply AtomNet® technology, which according to the company has found small molecule hits for more undruggable targets than any other AI drug discovery platform. Atomwise has raised more than $174 million in financing and attracted 250+ partners worldwide—including pharma giants that include Lilly and Bayer—carrying out 775 collaborations across more than 600 unique disease targets. Last month, Atomwise was one of 463 businesses worldwide to make Fast Company’s annual list of The World’s Most Innovative Companies.
On a smaller scale, Cambridge, MA-based Reverie Labs in February completed a $25 million Series A financing, whose proceeds will further develop the company’s computational kinase drug-discovery platform. Reverie’s pipeline includes in-house programs and those developed through a multi-target collaboration of undisclosed value with Roche and its Genentech subsidiary launched in July.
Seven-fold growth
The global market for AI-based drug discovery is projected to grow more than seven-fold, to $3.58 billion in 2027 from $473.4 million in 2019, according to GrandView Research.
Fueling that growth has been companies’ desire to cut down on the high costs of drug development in time and money—12 to 15 years, according to the NIH; and as much as $2.6 billion, according to a 2016 study by the Tufts Center for the Study of Drug Development published in Health Economics.
Nwankwo said that if she built an AI drug discovery company, three things would have to be different. “First, it’s not just that drug discovery is a problem. There are phases within it, and we need to build an AI platform or an ML platform for each of these problems, and that’s what you see at 1910 today.”
The second, she said, is 1910’s focus on basic biology, including assay biology, cell biology, and biochemistry, which has compelled the company to carry out that work itself rather than outsource it to CROs.
“At 1910, we say that ‘Biology is King’ because at the end of the day, all of these tools are to what end? You want to come up with a drug molecule that modulates a disease target, so the biology is actually where you go to get the verdict, as to whether or not all of this integration you’re doing with AI, with computation actually amounts to anything. I realized early on that, unlike many of the first-gen AI companies I’ve studied in this space, it was important for us to own our biology in-house.”
The third distinctive for 1910, Nwankwo said, was its computational focus: “We need to be very thoughtful in how we leverage the power of cloud computing, and that’s one of the things that excited me about having Microsoft lead the Series A round. Who better to help you build a world-class cloud computing infrastructure than the largest technology company in the world in the space?”
Target practice
1910 Genetics pursues targets in a variety of therapeutic areas. One is COVID-19, where at the onset of the pandemic last year, the company applied its AI tools by screening all known drug-like compounds for activity against SARS-CoV-2, identifying hit compounds after screening a billion-chemical library in less than six hours.
After synthesizing its target compounds to ensure that they were not cytotoxic, 1910 then confirmed them as actively inhibiting viral entry into Vero E6 cells engineered to overexpress TMPRSS2. 1910 focused on repurposed antivirals capable of inhibiting targets that included TMPRSS2 and another cell protease called furin. Nwankwo presented 1910 Genetics’ rapid AI-driven drug discovery effort during the first-ever NIH SARS-CoV-2 Virtual Summit, held in November.
Neuroscience was where the company said it identified its first target using its platforms, and where it has active candidates, without offering details. Nwankwo won’t discuss the status of her company’s other pipeline programs, which include aging, immunology, metabolic disorders, and oncology.
One of those areas is the focus of an ongoing collaboration with an unnamed pharma partner that reached out to 1910. “We haven’t been very aggressive with business development. That will come later.”
“We’re an early-stage, three-year-old company, so you can imagine that there are candidates in some of these, but certainly not anything that is in the clinic or even close to an IND or something like that track,” Nwankwo said. “We have not committed to what areas we want to define the company.”
Instead, 1910 Genetics has focused on developing two AI-based drug discovery engines consisting of several component platforms. One engine is used for small-molecule drugs; the other, for protein therapeutics. Both are intended to rapidly design therapeutics by integrating AI, big data, cloud computing, computational chemistry, quantum simulation, and biological automation.
“We want to play in that early discovery stage, where you have a target that you have very high conviction is playing a causative role in a disease. That’s where everything begins with 1910. It’s why the company’s called 1910!” Nwankwo said. “You begin by understanding a target really well, and you’re convinced without any doubt that if only I can modulate this target. I will cure this disease, or at least I will treat this disease. If you start from that vantage point, then it shouldn’t matter to you what therapeutic modality enables you to achieve that mission.”
1910 Genetics believes that some of the same AI architecture that it built to generate novel small molecules can also be used to generate novel protein sequences which can be peptides, fusions, or any number of biologics.
ELVIS sighting
The small-molecule discovery engine is named ELVIS™, whose pronunciation is derived from “ultra large-scale virtual screening” or ULVS, the approach to novel “hit” or target discovery applied by the first platform the company worked on.
ELVIS consists of three AI platforms:
- The gigascale ULVS platform renamed SUEDE™ for one of the real-life Elvis Presley’s first hits, “Blue Suede Shoes.” SUEDE is designed to identify promising hit compounds by virtually screening 14 billion molecules in less than six hours—compared to the six-month traditional pharmaceutical high throughput screening (HTS) that delivers hit compounds with success rates ranging from 0.01% to 1%, depending on target and concentration.
- BAGEL covers the hit-to-lead phase by designing novel scaffolds from molecular templates using neural networks. The generative chemistry platform generates de novo lead compounds using a hit compound as a template fast enough to cut the traditional hit to lead process from up to 24 months to two months, according to the company.
- CANDID is a drug property prediction platform for multi-parameter lead optimization, designed to ensure that a drug candidate balances potency with physical properties that ultimately determine its ideal absorption, distribution, metabolism, excretion and toxicity (ADMET) profile. CANDID is a lead-to-candidate platform built on graph neural networks (GNNs) that excel at tasks using the Quantitative structure-activity relationship (QSAR) computational modeling method.
“CANDID is also the place where molecules go to die,” Nwankwo quipped. “It’s like you need to give a candid assessment at that point, whether to further invest in this molecule because after this stage, we’re getting into later preclinical and things start to get very expensive.”
All three ELVIS AI platforms are integrated via cloud computing with 1910 Genetics’ in-house biological automation wet lab platform, which conducts in vitro biochemical and cell-based assays on the drug candidates and returns the wet lab data in a feedback loop to the AI platforms, while building a differentiated and proprietary data set.
“We have focused a lot on small molecule platforms to start out. And we’ve done lots of iterations on design-make-test there. We haven’t done as many iterations on the protein platforms, but that’s on our roadmap,” Nwankwo said.
In its quest against COVID-19, 1910 Genetics used ELVIS to design candidates that blocked SARS-CoV-2 entry into host cells. SUEDE screened a virtual library of one billion small molecules in hours and identified promising hit candidates, while BAGEL generated de novo small molecule drug candidates. After rapidly manufacturing both sets of molecules, 1910 Genetics tested them in-house using its automated wet lab platform, which validated the potency of two novel candidates in blocking SARS-CoV-2 entry into mammalian cells.
Recognizing Rosalind Franklin
For protein drug discovery, 1910 Genetics uses an engine named ROSALYND for Rosalind Franklin, who played a key role in the greatest biological discovery of the 20th century, the molecular structure of DNA.
“We stylized it with a Y because we had seen a couple of companies had used the Rosalind name for variety of things,” Nwankwo said. “I definitely wanted to recognize her. She’s one of the female scientists that have inspired me since I was a kid.”
Built on a proprietary data layer of 12 million 3D hydropathic interaction (HINT) maps, ROSALYND is used for de novo protein structure prediction, therapeutic protein design, and protein structure refinement. The HINT maps reflect the favorable and unfavorable substructures of predicted protein folds.
ROSALYND also includes a proprietary side-chain prediction platform built on advanced AI models, designed to drive protein refinement and de novo design pipelines. Nwankwo said ROSALYND is designed to offer a faster, more affordable and more accurate alternative to X-ray crystallography for 3D protein structure refinement, prediction, and de novo protein therapeutics design.
1910 has begun a tripling of its workforce, growing from roughly 10 staffers at the end of 2020, to 14 now, with plans to reach 30 by year’s end.
The company is not sharing much about its future financing plans, including if and when 1910 plans to go public. Nwankwo said she’s constantly fundraising. “All I’ll tell you is that all the cards are on the table,” she said.
“An additional amount of equity financing is absolutely on the table, but other forms of funding—partnerships with pharma co-development, research deals, joint ventures—there’s a lot of financial instruments that can get us to our goal of bringing medicines to patients faster,” Nwankwo added. “We will consider every one of them as they become appropriate and relevant.”
