Mon­tre­al AI start­up with a tout­ed ad­vi­sor li­cens­es its 'few-shot' plat­form to Re­pare for syn­thet­ic lethal­i­ty mol­e­cules

The rise of AI in drug dis­cov­ery has pre­sent­ed a buf­fet of tasty op­tions for drug­mak­ers to iden­ti­fy bet­ter mol­e­cules for hard-to-hit tar­gets. That promise is mixed, to say the least, but star­tups like Mon­tre­al’s Va­lence Dis­cov­ery think they’ve cracked the code — and at least one small biotech is con­vinced.

Va­lence, a re­cent­ly un­cloaked AI start­up spe­cial­iz­ing in deep learn­ing in drug dis­cov­ery, has li­censed its tech to Cana­da’s Re­pare Ther­a­peu­tics to iden­ti­fy ef­fec­tive syn­thet­ic lethal­i­ty mol­e­cules, the com­pa­nies said Wednes­day.

Yoshua Ben­gio

Va­lence’s plat­form seeks to over­come a crunch on avail­able train­ing da­ta sets by us­ing what it calls a “few-shot learn­ing” ap­proach that needs less start­ing da­ta to turn out valu­able mol­e­cules. Ac­cord­ing to CEO Daniel Co­hen, that ap­proach stems from re­search at the Mon­tre­al In­sti­tute for Learn­ing Al­go­rithms (Mi­la), the Mon­tre­al tech “ecosys­tem” spawned by Yoshua Ben­gio, a com­put­er sci­en­tist with a rep­u­ta­tion as a deep learn­ing pi­o­neer.

Ben­gio has joined Va­lence as an ad­vi­sor, and Co­hen thinks the team’s pro­pri­etary plat­form of­fers some­thing dif­fer­en­ti­at­ed over the many deep learn­ing play­ers on the mar­ket.

“The goal here is to help our part­ners very rapid­ly de­sign high-qual­i­ty drug can­di­dates that have been op­ti­mized for what­ev­er po­ten­cy, se­lec­tiv­i­ty, safe­ty prop­er­ties are rel­e­vant to that par­tic­u­lar drug dis­cov­ery pro­gram,” Co­hen said.

Va­lence didn’t di­vulge where Re­pare was look­ing to tar­get its mol­e­cules, Co­hen said.

While its work with Re­pare will tar­get syn­thet­ic lethal­i­ty drugs, Va­lence is tak­ing a broad ap­proach to po­ten­tial part­ner­ships — but the end goal is to of­fer its ser­vices to drug dis­cov­ery pro­grams where avail­able clin­i­cal da­ta are slim and tar­gets are hard to reach. Al­so, the com­pa­ny is work­ing to build a bet­ter “AI-gen­er­at­ed” mol­e­cule that has a high chance of suc­cess in the lab.

Daniel Co­hen

“AI-gen­er­at­ed mol­e­cules are of no val­ue to any­one if they can’t be eas­i­ly made in the lab,” Co­hen said. “Most AI sys­tems for drug de­sign to­day yield low-qual­i­ty mol­e­cules that are very dif­fi­cult to make. If you give an AI-gen­er­at­ed mol­e­cule to a med­i­c­i­nal chemist, maybe sev­en out of 10 times they’ll laugh at you. What we’ve done … is fo­cused on de­vel­op­ing new class­es of de­sign tech­nolo­gies that al­lows us to en­force a very high de­gree of qual­i­ty in­to our mol­e­cules.”

Va­lence is one of a num­ber of ear­ly play­ers that have skat­ed on the promise of an “AI-dis­cov­ered” mol­e­cule — a ques­tion­able claim that has whet­ted the in­dus­try’s ap­petite for a rev­o­lu­tion in the dis­cov­ery process.

AI has emerged as a cru­cial tool in both dis­cov­ery and pre­clin­i­cal R&D to mit­i­gate the bio­phar­ma in­dus­try’s abysmal at­tri­tion rate for new drugs, ef­fec­tive­ly giv­ing mol­e­cules a high­er chance of suc­cess, those com­pa­nies hope, be­fore they ever hit in vit­ro.

While some of those AI dis­cov­ery firms, like the UK’s Ex­sci­en­tia, have even­tu­al­ly mor­phed in­to full-on biotechs with their own in-house mol­e­cules, Va­lence is fo­cus­ing at the mo­ment on work­ing with its part­ners, Co­hen said.

Qual­i­ty Con­trol in Cell and Gene Ther­a­py – What’s Re­al­ly at Stake?

In early 2021, Bluebird Bio was forced to suspend clinical trials of its gene therapy for sickle cell disease after two patients in the trial developed cancer. As company scientists rushed to assess whether there was any causal link between the therapy and the cancer cases, Bluebird’s stock value plummeted – as did those of multiple other biopharma companies developing similar therapies.

While investigations concluded that the gene therapy was unlikely to have caused cancer, investors and the public may be more skittish regarding the safety of gene and cell therapies after this episode. This recent example highlights how delicate the fields of cell and gene therapy remain today, even as they show great promise.

Law pro­fes­sors call for FDA to dis­close all safe­ty and ef­fi­ca­cy da­ta for drugs

Back in early 2018 when Scott Gottlieb led the FDA, there was a moment when the agency seemed poised to release redacted complete response letters and other previously undisclosed data. But that initiative never gained steam.

Now, a growing chorus of researchers are finding that a dearth of public data on clinical trials and pharmaceuticals means industry and the FDA cannot be held accountable, two law professors from Yale and New York University write in an article published Wednesday in the California Law Review.

Endpoints Premium

Premium subscription required

Unlock this article along with other benefits by subscribing to one of our paid plans.

Novavax CEO Stanley Erck at the White House in 2020 (Andrew Harnik, AP Images)

As fears mount over J&J and As­traZeneca, No­vavax en­ters a shaky spot­light

As concerns rise around the J&J and AstraZeneca vaccines, global attention is increasingly turning to the little, 33-year-old, productless, bankruptcy-flirting biotech that could: Novavax.

In the now 16-month race to develop and deploy Covid-19 vaccines, Novavax has at times seemed like the pandemic’s most unsuspecting frontrunner and at times like an overhyped also-ran. Although they started the pandemic with only enough cash to last 6 months, they leveraged old connections and believers into $2 billion and emerged last summer with data experts said surpassed Pfizer and Moderna. They unveiled plans to quickly scale to 2 billion doses. Then they couldn’t even make enough material to run their US trial and watched four other companies beat them to the finish line.

FDA of­fers scathing re­view of Emer­gent plan­t's san­i­tary con­di­tions, em­ploy­ee train­ing af­ter halt­ing pro­duc­tion

The FDA wrapped up its inspection of Emergent’s troubled vaccine manufacturing plant in Baltimore on Tuesday, after halting production there on Monday. By Wednesday morning, the agency already released a series of scathing observations on the cross contamination, sanitary issues and lack of staff training that caused the contract manufacturer to dispose of millions of AstraZeneca and J&J vaccine doses.

Brad Bolzon (Versant)

Ver­sant pulls the wraps off of near­ly $1B in 3 new funds out to build the next fleet of biotech star­tups. And this new gen­er­a­tion is built for speed

Brad Bolzon has an apology to offer by way of introducing a set of 3 new funds that together pack a $950 million wallop in new biotech creation and growth.

“I want to apologize,” says the Versant chairman and managing partner, laughing a little in the intro, “that we don’t have anything fancy or flashy to tell you about our new fund. Same team, around the same amount of capital, same investment strategy. If it ain’t broke, don’t fix it.”

But then there’s the flip side, where everything has changed. Or at least speeded into a relative blur. Here’s Bolzon:

Endpoints Premium

Premium subscription required

Unlock this article along with other benefits by subscribing to one of our paid plans.

Sen. Patty Murray (D-WA) (Graeme Sloan/Sipa USA/Sipa via AP Images)

Sen­a­tors to NIH: Do more to pro­tect US bio­med­ical re­search from for­eign in­flu­ence

Although Thursday’s Senate health committee hearing was focused on how foreign countries and adversaries might be trying to steal or negatively influence biomedical research in the US, the only country mentioned by the senators and expert witnesses was China.

Committee chair Patty Murray (D-WA) made clear in her opening remarks that the US cannot “let the few instances of bad actors” overshadow the hard work of the many immigrant researchers in the US, many of which have won Nobel prizes for their work. But she also said, “There is more the NIH can be doing here.”

Noubar Afeyan (Sebastien Micke/Paris Match/Contour by Getty Images)

As Mod­er­na rose, Flag­ship cashed in for $1.4B — with a lot more wealth still re­main­ing

For nearly a decade, Flagship poured record-setting levels of cash into Moderna, even as they faced setbacks on early programs and skeptics wondered whether the company’s science could ever match its hype.

Now that the science has delivered, Flagship is cashing in.

Over the last 13 months, since the World Health Organization declared a pandemic, Flagship has sold off Moderna shares worth $1.4 billion. The sales, first reported by Forbes, came as the Cambridge biotech’s shares soared from just under $20 per share on Jan. 3, 2020, to $169.50 when markets opened Thursday.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 107,500+ biopharma pros reading Endpoints daily — and it's free.

Paul Wotton, Obsidian CEO

Ver­tex adds a new gene edit­ing part­ner, pay­ing Ob­sid­i­an $75M for con­trol­lable CRISPR sys­tem

When Paul Wotton became CEO of Obsidian Therapeutics two years ago, he asked his team to explore where they could apply the company’s technology beyond cancer.

Launched by Michael Gilman in 2017, the GV-backed company had been working on a way of controlling cell therapies after they’re administered to a patient, allowing doctors to modulate the potency of CAR-T and other drugs as you might volume on a television set. Wotton wanted to them look beyond cancer, into areas such as gene editing.

Jenny Rooke (Genoa Ventures)

Ear­ly Zymer­gen in­vestor Jen­ny Rooke re­flects on 'chimeras' in biotech, what it takes to spot a $500M gem

When Jenny Rooke first heard of Zymergen back in 2014, she knew she was looking at something different and exciting. The Emeryville, CA biotech held the promise of blending biology and technology to solve a huge unmet need for cost-effective chemicals — of all things — and a stellar founding team to boot.

But back then, West Coast venture capitalists didn’t see in Zymergen the one thing they were looking for in a winning biotech: therapeutic potential. Rooke, however, saw an opportunity and made her bets. Seven years later, that bet is paying off in a big way.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 107,500+ biopharma pros reading Endpoints daily — and it's free.