David Roblin (Relation)

Ex­clu­sive: Pfiz­er vet grabs $25M seed round to 'in­te­grate' ma­chine learn­ing, ge­nomics tech in drug R&D

Hav­ing been in­volved in a num­ber of at­tri­tion task force ef­forts dur­ing his tenure as Eu­ro­pean R&D chief at Pfiz­er, David Rob­lin tells me there’s a quote that res­onates with him, ut­tered 20 years ago by South African bi­ol­o­gist Syd­ney Bren­ner when he ac­cept­ed a No­bel Prize.

“We are drown­ing in a sea of da­ta but thirsty for knowl­edge,” is how Rob­lin re­mem­bers it.

But the way those da­ta have been used to date — by drug dis­cov­ery sci­en­tists try­ing to val­i­date their ideas — hasn’t re­al­ly im­proved the rate at which pre­clin­i­cal com­pounds be­come mar­ket­ed drugs, he reck­ons. The po­ten­tial he sees for ma­chine learn­ing and new tech­nolo­gies like sin­gle-cell tran­scrip­tomics to change that is the rea­son why, af­ter eight years as chief of sci­en­tif­ic trans­la­tion at Lon­don’s Fran­cis Crick In­sti­tute, he’s mov­ing down the road to Re­la­tion Ther­a­peu­tics, which has just closed a $25 mil­lion seed round.

“We’re not go­ing to sit here and tell you that ma­chine learn­ing is the sin­gle an­swer and every­thing will be de­signed in the ma­chine and that ma­chine will send a file to the FDA,” he said. “The key ques­tion for a com­pa­ny like ours is, where do you best de­ploy ma­chine learn­ing that af­fords max­i­mum im­pact? Where do you use tran­scrip­tomics? And where do you see the link­age?”

Rob­lin, the CEO, is joined in the C-suite by Lind­say Ed­wards, CTO and pres­i­dent of plat­form. The for­mer head of AI for res­pi­ra­to­ry and im­munol­o­gy at As­traZeneca, Ed­wards al­so led one of GSK’s first da­ta sci­ence groups be­fore that.

Com­ing out of stealth at a time the idea of de­ploy­ing AI/ML in drug R&D has al­ready gone through a cou­ple cy­cles of hype and bust, Re­la­tion’s ap­proach will be ground­ed in ac­tive learn­ing, where ma­chine learn­ing sys­tems will go through the da­ta gen­er­at­ed in its wet and in sil­i­co labs, and then tell the sci­en­tists what new ex­per­i­ments to run for it to come up with new in­sight.

Specif­i­cal­ly, Ed­wards not­ed, the com­pa­ny will be us­ing a frame­work called ac­tive-graph ma­chine learn­ing, or Meta­graph, us­ing the con­cept of graphs as a back­bone so as to map out the com­plex re­la­tion­ships be­tween genes, pro­teins and drugs.

“What Lind­say is do­ing, large­ly, is tak­ing al­go­rithms, ideas that have been de­ployed else­where, par­tic­u­lar­ly in con­sumer so­cial net­work­ing, and turn­ing them to the pur­pose of gene vari­ant bi­ol­o­gy, iden­ti­fi­ca­tion,” Rob­lin said, adding, “So in that sense, I’m quite hap­py to say like, a lot of that stuff is not ab­solute­ly break­through. What we’re go­ing to be do­ing is cre­at­ing a plat­form of in­te­gra­tion which is sec­ond to none, es­sen­tial­ly.”

Based in the Knowl­edge Quar­ter in Lon­don — sur­round­ed by UCL, King’s and Fran­cis Crick as well as sev­er­al hos­pi­tals — Re­la­tion is look­ing to crunch ge­nom­ic da­ta on the in­di­vid­ual cell lev­el with sam­ples pro­vid­ed by neigh­bor­ing biobanks.

As its first big project, the team is de­vel­op­ing a cell at­las of the bone, an area where sig­nif­i­cant gaps in bi­o­log­i­cal un­der­stand­ing still ex­ist, ac­cord­ing to Rob­lin. It is im­por­tant, in his words, to ful­ly grasp the trans­la­tion­al route and en­sure the cell mod­els used in Re­la­tion’s labs are “as close to re­ca­pit­u­la­tion of dis­ease as pos­si­ble.”

“Mak­ing sure you work on the right things is a re­al­ly crit­i­cal prob­lem,” Ed­wards said.

Re­la­tion will al­so get help from Nvidia, which is giv­ing it ac­cess to the NVIDIA Cam­bridge-1 GPU su­per­com­put­er, which helps opens up new pos­si­bil­i­ties for look­ing at large stretch­es of DNA.

“Every­body runs in­to a com­pute lim­i­ta­tion, and that ba­si­cal­ly sets the width of the amount of DNA that you can look at in a sin­gle go,” he said. “So ac­cess to Cam­bridge lab, we ex­pect will give us some kind of break­through re­cep­tive fields with our mod­els.”

With “re­al­ly good R&D peo­ple” at the cen­ter of it all, Rob­lin said Re­la­tion will aim to grow from 15 to 40 staffers with the new cash. The com­pa­ny has al­so put to­geth­er a hefty sci­en­tif­ic ad­vi­so­ry board, in­clud­ing Michael Bron­stein (Ox­ford pro­fes­sor and head of graph AI at Twit­ter), as well as Alex Shalek and Car­o­line Uh­ler of the Broad In­sti­tute.

DCVC and Mag­net­ic Ven­tures co-led the fi­nanc­ing, with par­tic­i­pa­tion from Khosla Ven­tures, OMERS Ven­tures, and first­minute Cap­i­tal, plus a num­ber of in­di­vid­ual in­vestors.

Up­dat­ed: FDA re­mains silent on or­phan drug ex­clu­siv­i­ty af­ter last year's court loss

Since losing a controversial court case over orphan drug exclusivity last year, the FDA’s Office of Orphan Products Development has remained entirely silent on orphan exclusivity for any product approved since last November, leaving many sponsors in limbo on what to expect.

That silence means that for more than 70 orphan-designated indications for more than 60 products, OOPD has issued no public determination on the seven-year orphan exclusivity in the Orange Book, and no new listings of orphan exclusivity appear in OOPD’s searchable database, as highlighted recently by George O’Brien, a partner in Mayer Brown’s Washington, DC office.

Big week for Alzheimer’s da­ta; As­traZeneca buys cell ther­a­py start­up; Dig­i­tal ther­a­peu­tics hits a pay­er wall; and more

Welcome back to Endpoints Weekly, your review of the week’s top biopharma headlines. Want this in your inbox every Saturday morning? Current Endpoints readers can visit their reader profile to add Endpoints Weekly. New to Endpoints? Sign up here.

You may start to notice more stories exclusively available to Premium subscribers. This week alone, paid subscribers can read our in-depth reporting on Alzheimer’s data, digital therapeutics and Allogene’s cell therapy for solid tumors, as well as scoops on Twitter ads and Catalent. With your support, we can keep growing our team and spend more time on quality work. We have both individual and company plans available — check them out to unlock the full Endpoints experience.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 154,000+ biopharma pros reading Endpoints daily — and it's free.

Illustration: Assistant Editor Kathy Wong for Endpoints News

As mon­ey pours in­to dig­i­tal ther­a­peu­tics, in­sur­ance cov­er­age crawls



Talk therapy didn’t help Lily with attention deficit hyperactivity disorder, or ADHD. But a video game did.

As the 10-year-old zooms through icy waters and targets flying creatures on the snow-capped planet Frigidus, she builds attention skills, thanks to Akili Interactive Labs’ video game EndeavorRx. She’s now less anxious and scattered, allowing her to stay on a low dose of ADHD medication, according to her mom Violet Vu.

Endpoints Premium

Premium subscription required

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

Chris Kim, Liminatus Pharma CEO

A fledg­ling biotech goes SPAC route to bankroll can­cer vac­cine, CAR-T and CD47

A relatively unknown clinical-stage biotech — backed by a Korean lighting company and focused on a cancer vaccine out of a Thomas Jefferson University lab — is headed to Nasdaq via the blank check route.

Liminatus Pharma will get about $316 million in proceeds from the SPAC combination to fund its ongoing Phase IIa study of a cancer vaccine, bring a CAR-T therapy into the clinic and prep a CD47 immune checkpoint inhibitor for human trials, the company said this week.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 154,000+ biopharma pros reading Endpoints daily — and it's free.

Am­gen, years be­hind ri­vals, says PhI obe­si­ty drug shows dura­bil­i­ty signs

While NBC ran “The Biggest Loser” for 17 seasons, deemed toxic by critics for the reality show’s punishing exercise and diet upheavals, researchers in pharmaceutical labs have been attempting to create prescription drugs that induce weight loss — and one pharma betting it can require less frequent dosing is out with a new crop of data.

Amgen was relatively late to the game compared to its approved competitor Novo Nordisk and green light-approaching rival Eli Lilly. But early data suggested Amgen’s AMG 133 led to a 14.5% weight reduction in the first few months of dosing, buoying shares earlier this fall, and now the California pharma is out with its first batch of durability data showing that figure fell slightly to 11.2% about 150 days after the last dose. Amgen presented at the 20th World Congress on Insulin Resistance, Diabetes & Cardiovascular Disease on Saturday afternoon.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 154,000+ biopharma pros reading Endpoints daily — and it's free.

Eli Lil­ly’s Alzheimer’s drug clears more amy­loid ear­ly than Aduhelm in first-ever head-to-head. Will it mat­ter?

Ahead of the FDA’s decision on Eli Lilly’s Alzheimer’s drug donanemab in February, the Big Pharma is dropping a first cut of data from one of the more interesting trials — but less important in a regulatory sense — at an Alzheimer’s conference in San Francisco.

In the unblinded 148-person study, Eli Lilly pitted its drug against Aduhelm, Biogen’s drug that won FDA approval but lost Medicare coverage outside of clinical trials. Notably, the study didn’t look at clinical outcomes, but rather the clearance of amyloid, a protein whose buildup is associated with Alzheimer’s disease, in the brain.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 154,000+ biopharma pros reading Endpoints daily — and it's free.

Uğur Şahin, BioNTech CEO (ddp images/Sipa USA/Sipa via AP Images)

BioN­Tech bets on dif­fi­cult STING field via small mol­e­cule pact with a Pol­ish biotech

BioNTech is beefing up its relatively thin small molecule pipeline by adding weight to a clinically difficult corner of oncology R&D: STING agonists. To do so, BioNTech is teaming up with a 15-year-old Polish biotech and doling out €40 million, about $41.5 million, to start.

The deal is broken into two parts: First, BioNTech obtains an exclusive global license to develop and market Ryvu Therapeutics’ STING agonist portfolio as small molecules, whether alone or in combination with other agents.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 154,000+ biopharma pros reading Endpoints daily — and it's free.

Bay­er starts work on $43M+ ex­pan­sion of OTC man­u­fac­tur­ing site in Penn­syl­va­nia

German pharma giant Bayer will be looking to make a significant investment into one of its US plants that produces over-the-counter drugs.

Bayer announced that it will spend $43.6 million to expand its facility in Myerstown, PA, a small town east of Harrisburg. Bayer plans to increase the site by 70,000 square feet and will have room for the installation of eight packaging lines and an area to install rooftop solar panels. The project is expected to be completed by 2025 and will add around 50 to 75 jobs.

US month­ly costs for biosim­i­lars 'sub­stan­tial­ly high­er' than Ger­many or Switzer­land, JA­MA re­search finds

As the global biologics market is expected to hit nearly the half-trillion-dollar mark this year, new JAMA research points to the importance of timely biosimilar entry, particularly as fewer biosimilars are entering the US than in Europe, and as monthly treatment costs for biosimilars were “substantially higher” in the US compared with Germany and Switzerland.

Among the three countries, biosimilar market share at launch was highest in Germany, but increased at the fastest rate in the US, the authors from the University of Zurich’s Institute of Law wrote in JAMA Network Open today.