Bruce Beutel

If an AI drug discovery platform doesn’t produce multiple drug candidates, it’s still a demo…🔬🤷‍♂️ 📄 In this new piece for R&D World, Brian K. Buntz interviews Alex Zhavoronkov, CEO of Insilico Medicine, in a direct and fact-driven discussion on the one thing AI drug discovery desperately needs: benchmarks 📊. Titled “Compete, don’t complain: Insilico measures AI drug discovery by benchmarks, not talk” the article offers a grounded view of what it takes to move beyond AI theory and actually build a company that delivers validated drug candidates, at speed and scale. As an early investor in Insilico Medicine and through LongeVC, I’ve seen first-hand how the team built what many in the space only pitch: a full-stack, AI-native engine that helps turn in silico ideas into real drug candidates, quickly and repeatedly. 🚀 A few strategic aspects I think made Insilico what it is today: ✅ Tech stack as infrastructure, not a collection of tools: Insilico’s PharmaAI platform integrates target discovery (PandaOmics), molecular generation (Chemistry42), and clinical trial prediction (inClinico) into a unified pipeline, not as siloed features, but as a single operating system for drug R&D. ✅ Hardware and automation are not afterthoughts: The company’s Suzhou-based “Life Star” lab uses fully automated execution, including a bipedal humanoid robot, to physically validate AI-generated compounds at scale. It is certainly one of the most tightly integrated digital-to-wet-lab pipelines in the industry. ✅ Global structure built for speed: AI R&D in Montreal and Abu Dhabi; synthesis, preclinical, and early clinical work in Asia. Thinking strategically about geography is crucial in the modern, politically, economically, and regulatorily-complex world, and it’s how Insilico achieves operational efficiency both in R&D (as short as 9 months from target to development candidate) and business growth. ✅ Clear long-term thesis on aging: The “longevity vault” now holds 80+ dual-purpose targets that map to both aging and disease. The strategy is disciplined: get there through disease first, build clinical and regulatory momentum, and be ready when the time is right 🎯. 💭 “You don’t want to be assembling iPhones — you need to figure out how to develop the next iPhone if China is already assembling them efficiently,” Alex says, drawing a parallel between biotech innovation and how companies compete in a global system built on execution speed and scale If you're following the intersection of AI, biotech, and translational longevity, this is a company (and article) to study closely. Link in the comments 👇

139

8 Comments

When sequencing becomes that accessible, entirely new categories of research and product become viable. Element Biosciences just pushed the benchmark forward and San Diego stays at the center of it. Full breakdown with Fred Grier https://lnkd.in/gAY7jYaM

8

Recent commentary on the situation of REGENXBIO’s CRL got me thinking a lot about the evidentiary standard being applied to surrogate endpoints and the consistency of that standard. For background to those unfamiliar: FDA issued a CRL (Complete Response Letter) to REGENXBIO (RGX) for their gene therapy RGX-121 treating Hunter syndrome -  for my non-healthcare affiliated LinkedIn folks, a CRL is basically a communication letter from the FDA that signals a drug is not ready for approval and recommends corrective actions. RGX took a path for accelerated approval, which can utilize a surrogate endpoint as a clinical biomarker – one that, according to the FDA, needs to be "reasonably likely to predict clinical benefit." I could write an essay alone about the biomarker of choice here, CSF HS D2S6, but in short REGENXBIO proposed CSF levels of D2S6 specifically as their surrogate endpoint, arguing it correlates with neurocognitive outcomes. D2S6 is one of four subunits of heparan sulfate that accumulates in the brains of Hunter syndrome patients and drives their neurological decline. The FDA pushed back that the D2S6 subunit in specific lacks external literature support and wasn't established in standard clinical practice as a validated biomarker for the disease, ultimately not clearing the bar of "reasonably likely to predict clinical benefit." Reflecting on this tension got me thinking about how our field is sometimes navigating two schools of thought for resolving scientific, evidentiary disputes: fixed and rigorous or contextually flexible (both of which are internally coherent). What I mean by this is that we can defend a treatment with scientific/statistical burden of proof (i.e. D2S6 reducing enzymatic burden and restoring neurological function); however, where do we scientifically draw the line of "reasonably likely to predict clinical benefit," especially in the stress-tested conditions of accelerated approval? I’ve heard some argue that it is like the legal preponderance of evidence standard, or "more likely than not," following a different risk tolerance to predict clinical benefit. However, other critics of accelerated approval have also argued that it should meet a standard more closely related to “beyond reasonable doubt.” But the two positions are interesting because, unlike the legal framework, drug approvals have bidirectional tension potentially calling into question "beyond reasonable doubt" (I’m speaking generally here, put the pitchforks away). Our field, especially rare diseases, often faces the dilemma of approving an ineffective or unsafe therapy that harms patients or not approving an effective therapy for a fatal disease which also harms patients. This isn’t easily solvable for a reason, but it’s brought about great conversations regarding biomarkers as a whole and how we apply evidence frameworks to accelerated approvals and rare disease.

37

6 Comments

The Tao of Biotech: Discovery, Finance, Litigation. One of the most fascinating biotech IP stories of the past two decades started with a small Canadian company few people outside RNA therapeutics circles remember: Tekmira Pharmaceuticals, originally Inex, co-founded by legendary formulator Pieter Cullis. Tekmira’s scientists were early pioneers in lipid nanoparticle (LNP) delivery, developing ways to package fragile RNAi inside lipid particles as an in vivo drug delivery system. One of the early validation moments for the technology came through initial partnership with Alnylam, then acrimonious lawsuits followed over patent breach, and finally, $65M settlement from Alnylam, with continued drug collaboration. Business is business, after all and these activities ultimately led to the first approved RNAi drug, Onpattro. Clearly, Tekmira was the undisputed leader in LNP formulations. Tekmira's story took a major turn in 2015, when they merged with OnCore BioPharma, a hepatitis B–focused biotech founded by several Pharmasset veterans. Following the merger, the company rebranded as Arbutus Biopharma, signaling a shift toward building a pipeline of HBV therapies while continuing to hold a valuable portfolio of LNP delivery patents. Those patents would become even more important a few years later. In 2018, Arbutus partnered with Roivant Sciences, founded by Vivek Ramaswamy (the young billionaire hedge fund genius and later presidential candidate, and now candidate for Ohio governor....) to create Genevant Sciences — a platform company designed to license LNP delivery technology across the rapidly growing RNA therapeutics ecosystem. Thus, Arbutus retained ownership of foundational LNP patents, Genevant became the licensing and commercialization engine, and Roivant became a major owner and platform partner, ready to monetize any upside. In effect, Genevant positioned itself as a toll booth for RNA delivery technologies, which in 2018, was two years ahead of jackpot. During the COVID, $100's of billions were generated in vaccines using LNPs. Arbutus and Genevant bitterly asserted that Moderna’s mRNA vaccines used technology covered by their patents. The resulting global litigation has now culminated in a settlement reportedly worth up to $2.25B. This pattern is strangely reminiscent of earlier Tekmira / Alnylam litigation. It is a common theme that securing a reliable drug delivery system is often overlooked during the drug discovery, because industry is so heavily focused on the developing the active pharmaceutical ingredient (note Cullis was overlooked for Nobel Prize). Moderna market cap is $20 Bn, of which now must reflect a potential penalty loss of up to $2 Bn. Perhaps this is a wake up call to future endeavors in drug discovery campaigns to include delivery systems (and their licensing) as a necessary partner to making therapies. https://lnkd.in/emkzvmN6

78

10 Comments

Looking to streamline data flow, simplify QC, or boost method robustness? SciY's latest roundup brings together a peer‑reviewed PAT article, an app note, a tech‑transfer blog, and an on‑demand webinar on pharma data strategies. Read the full selection 👇 #SciY

9

🚨 Another signal that gene therapy is far from slowing down. Kriya Therapeutics, Inc. Therapeutics just announced a $313M financing round—the 6th largest private biotech raise of the year. In today’s cautious funding climate, this kind of mega-round sends important signals about where the field is headed: 1️⃣ Investor confidence is rebounding. Despite questions about the commercial success of approved gene therapies, investors are clearly betting big on platforms with breadth and scalability. Kriya’s raise underscores that conviction. 2️⃣ Ophthalmology and metabolic disorders are rising to the forefront. Eye diseases like geographic atrophy offer localized delivery and clear clinical endpoints, while metabolic diseases (MASH, type 1 diabetes) represent expansion into larger, chronic disease markets. The field is no longer limited to ultra-rare conditions. 3️⃣ Scale is the new differentiator. In a capital-constrained environment, only a handful of well-funded companies can advance multiple programs in parallel. Kriya’s war chest positions it to be one of those platform players shaping the next chapter of gene therapy. ✨ For the gene therapy sector, this is a reminder: capital is available, but it’s flowing toward companies that show investors they can execute across programs and markets—not just single assets. What do you think—will we see more mega-rounds like this in 2025, or will capital consolidate even further into just a few players? #genetherapy #advancedtherapies #biotech #lifesciences #funding #313m https://lnkd.in/eEHS4K-u

162

3 Comments