DeepMind CEO Demis Hassabis

Al­pha­bet kick­starts new AI drug dis­cov­ery out­fit with lessons learned from Deep­Mind pro­tein break­throughs

Last sum­mer marked a ma­jor break­through in drug dis­cov­ery when Deep­Mind, a pre­dic­tive mod­el­ing start­up from Google par­ent com­pa­ny Al­pha­bet, of­fered the most ac­cu­rate pic­ture yet of the “pro­tein fold­ing” prob­lem. The Al­pha­bet team is now prop­ping up a unit fo­cused sole­ly on drug dis­cov­ery, and it will look to lever­age lessons learned from Deep­Mind’s ex­am­ple.

Al­pha­bet has launched Iso­mor­phic Labs, a Lon­don-based drug dis­cov­ery start­up lever­ag­ing the com­pa­ny’s AI and ma­chine learn­ing work, and lessons from Deep­Mind’s Al­phaFold break­throughs, CEO Demis Has­s­abis said in a blog post Thurs­day.

You’ll know Has­s­abis as the CEO of Al­pha­bet’s Deep­Mind, the com­pa­ny that of­fered the most ac­cu­rate vi­su­al mod­el yet of “pro­tein fold­ing,” a no­to­ri­ous chal­lenge that had elud­ed re­searchers for more than 50 years.

Has­s­abis will re­main in his role at Deep­Mind while lead­ing what he called the “first phase” of growth at Iso­mor­phic, in­clud­ing flesh­ing out the com­pa­ny’s strate­gic vi­sion and hir­ing on key staff. Here’s what he had to say about the com­pa­ny’s mis­sion:

We are at an ex­cit­ing mo­ment in his­to­ry now where (AI and ma­chine learn­ing) are be­com­ing pow­er­ful and so­phis­ti­cat­ed enough to be ap­plied to re­al-world prob­lems in­clud­ing sci­en­tif­ic dis­cov­ery it­self. One of the most im­por­tant ap­pli­ca­tions of AI that I can think of is in the field of bi­o­log­i­cal and med­ical re­search, and it is an area I have been pas­sion­ate about ad­dress­ing for many years. Now the time is right to push this for­ward at pace, and with the ded­i­cat­ed fo­cus and re­sources that Iso­mor­phic Labs will bring.

While there will be some crossover in terms of Iso­mor­phic and Deep­Mind’s ex­per­tise, a com­pa­ny spokesper­son said the units would op­er­ate in­de­pen­dent­ly with the for­mer fo­cus­ing on cre­at­ing a “com­pu­ta­tion­al plat­form to un­der­stand bi­o­log­i­cal sys­tems from first prin­ci­ples to dis­cov­er new ways to treat dis­ease.”

Iso­mor­phic is, of course, not the first com­pa­ny look­ing to lever­age break­throughs in AI to change the game of drug de­vel­op­ment, a chron­i­cal­ly tricky en­deav­or with very low rates of suc­cess. Com­pa­nies in the vein of in­sitro and Ex­sci­en­tia have claimed own­er­ship of “AI-dis­cov­ered” mol­e­cules — with some de­bate over who fin­ished first — but those re­cent re­sults still have a long way to go to prove them­selves in the clin­ic.

But com­ing from the team be­hind Al­phaFold, Iso­mor­phic is like­ly to catch a lot of eye­balls as a rel­a­tive­ly late en­trant in­to the field. In Ju­ly, Deep­Mind an­nounced it would open its pro­tein data­base, culled from Al­phaFold, to sci­ence in what looked like a ma­jor win for drug de­vel­op­ers. Al­phaFold’s pre­dic­tive mod­el­ing sys­tem us­es pro­tein se­quences to of­fer vi­su­al mod­els for re­searchers, po­ten­tial­ly shed­ding new light on drug tar­gets, ex­perts say.

While there ap­pears to be some crossover in terms of what Iso­mor­phic and Deep­Mind are do­ing on this front, Al­pha­bet said Al­phaFold will still be work­ing in this space sep­a­rate­ly, with po­ten­tial learn­ings shared across groups. That’s a good thing be­cause ac­tu­al­ly mak­ing mean­ing­ful head­way in drug dis­cov­ery will like­ly take as many shots on goal as pos­si­ble.

“Bi­ol­o­gy is like­ly far too com­plex and messy to ever be en­cap­su­lat­ed as a sim­ple set of neat math­e­mat­i­cal equa­tions. But just as math­e­mat­ics turned out to be the right de­scrip­tion lan­guage for physics, bi­ol­o­gy may turn out to be the per­fect type of regime for the ap­pli­ca­tion of AI,” he wrote.

We’ll see.

Alexander Lefterov/Endpoints News

A new can­cer im­munother­a­py brings cau­tious hope for a field long await­ing the next big break­through

Bob Seibert sat silent across from his daughter at their favorite Spanish restaurant near his home in Charleston County, SC, their paella growing cold as he read through all the places in his body doctors found tumors.

He had texted his wife, a pediatric intensive care nurse, when he got the alert that his online chart was ready. Although he saw immediately it was bad, many of the terms — peritoneal, right iliac — were inscrutable. But she was five hours downstate, at a loud group dinner the night before another daughter’s cheer competition.

Endpoints Premium

Premium subscription required

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

Bahija Jallal, Immunocore CEO

BREAK­ING: FDA hur­ries up a quick ap­proval for the world's first TCR -- af­ter a 14-year R&D trek

Over the 14 years since Immunocore was spun out of MediGene in a quest to develop a gamechanging cancer med, the biotech has raised record sums and undergone a major shakeup on a long roller coaster ride of valuations for investors. But they survived and thrived and today they’re popping the champagne corks to celebrate an FDA approval of their first TCR drug.

Immunocore flagged the FDA’s green light for tebentafusp Wednesday morning by highlighting a series of firsts.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 129,700+ biopharma pros reading Endpoints daily — and it's free.

Dominic Borie, Kyverna Therapeutics CEO

Well-con­nect­ed, Gilead-backed biotech gets an­oth­er stack of cash to pur­sue CAR-Ts for au­toim­mune dis­ease

Almost exactly two years after its debut at the 2020 JP Morgan confab — and on the heels of a new partnership with the gene editing experts at Intellia — a Gilead-backed, autoimmune disease-focused startup has returned to the well with a clearer outline of just what it plans to do with its CAR-T platform.

Kyverna brought in $85 million in its oversubscribed Series B, the company announced Wednesday. Northpond Ventures led the round, and Westlake Village BioPartners, Vida Ventures, Gilead and Intellia all contributed as well.

Days af­ter Gilead yanks PI3K drug, In­cyte with­draws NDA for its own PI3K — say­ing con­fir­ma­to­ry tri­als would take too long

The FDA’s intensifying scrutiny on accelerated approvals isn’t just putting pressure on drugmakers with marketed products. It is also subtly reshaping the regulatory dynamics.

Case in point: Incyte announced late Tuesday that it has made the “business decision” to withdraw an NDA for parsaclisib, its oral PI3Kδ inhibitor, after deciding that running the confirmatory studies the agency was asking for to support an accelerated approval wouldn’t be worth it.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 129,700+ biopharma pros reading Endpoints daily — and it's free.

Dan O'Day (Getty Images)

In a set­back, FDA or­ders Gilead to hit the brakes on their late-stage, $5B can­cer play

Gilead’s $5 billion drug magrolimab has run into a serious setback.

The FDA ordered Gilead to halt enrollment on their studies of the drug in combination with azacitidine after investigators reports revealed an “apparent imbalance” in the suspected unexpected serious adverse reactions between study arms. And the halt is raising questions about Gilead’s plans for a quick pitch to regulators.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 129,700+ biopharma pros reading Endpoints daily — and it's free.

Graphic: Alexander Lefterov for Endpoints News

Small biotechs with big drug am­bi­tions threat­en to up­end the tra­di­tion­al drug launch play­book

Of the countless decisions Vlad Coric had to make as Biohaven’s CEO over the past seven years, there was one that felt particularly nerve-wracking: Instead of selling to a Big Pharma, the company decided it would commercialize its migraine drug itself.

“I remember some investors yelling and pounding on the table like, you can’t do this. What are you thinking? You’re going to get crushed by AbbVie,” he recalled.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 129,700+ biopharma pros reading Endpoints daily — and it's free.

Mar­ket­ingRx roundup: Pfiz­er de­buts Pre­vnar 20 TV ads; Lil­ly gets first FDA 2022 pro­mo slap down let­ter

Pfizer debuted its first TV ad for its Prevnar 20 next-generation pneumococcal pneumonia vaccine. In the 60-second spot, several people (actor portrayals) with their ages listed as 65 or older are shown walking into a clinic as they turn to say they’re getting vaccinated with Prevnar 20 because they’re at risk.

The update to Pfizer’s blockbuster Prevnar 13 vaccine was approved in June, and as its name suggests is a vaccine for 20 serotypes — the original 13 plus seven more that cause pneumococcal disease. Pfizer used to spend heavily on TV ads to promote Prevnar 13 in 2018 and 2019 but cut back its TV budgets in the past two fall and winter seasonal spending cycles. Prevnar had been Pfizer’s top-selling drug, notching sales of just under $6 billion in 2020, and was the world’s top-selling vaccine before the Covid-19 vaccines came to market last year.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 129,700+ biopharma pros reading Endpoints daily — and it's free.

Roy Baynes, Merck

FDA bats back Mer­ck’s ‘pipeline in a prod­uct,’ de­mands more ef­fi­ca­cy da­ta

Despite some heavy blowback from analysts, Merck execs maintained an upbeat attitude about the market potential of its chronic cough drug gefapixant. But the confidence may be fading somewhat today as Merck puts out news that the FDA is handing back its application with a CRL.

Dubbed by Merck’s development chief Roy Baynes as a “pipeline in a product” with a variety of potential uses, Merck had fielded positive late-stage data demonstrating the drug’s ability to combat chronic cough. The drug dramatically reduced chronic cough in Phase III, but so did placebo, leaving Merck’s research team with a marginal success on the p-value side of the equation.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 129,700+ biopharma pros reading Endpoints daily — and it's free.

Florida Gov. Ron DeSantis (AP Photo/Wilfredo Lee, File)

Opin­ion: Flori­da is so mAb crazy, Ron De­San­tis wants to use mAbs that don't work

Florida Gov. Ron DeSantis is trying so hard to politicize the FDA and demonize the federal government that he entered into an alternate universe on Monday evening in describing a recent FDA action to restrict the use of two monoclonal antibody, or mAb, treatments for Covid-19 that don’t work against Omicron.

Without further ado, let’s break down his statement from last night, line by line, adjective by adjective.