Where Tech Meets Bio (Substack Newsletter)

Looking forward to the report by Andrii Buvailo, Ph.D. . But I do not necessarily agree that data is most important. The ability to rapidly validate is most important.Second most important success factor is community feedback and validation. Your tools need to work in the hands of the other companies. And with 22 PCCs and 10 clinical-stage assets in 4 years, I can make certain claims from time to time. We now have over $3 Trillion dollars worth of data at Insilico (based on grants we monitor), probably the biggest data repository in the world. SE of this data is no longer available publicly so if you were not collecting it in 2015, you don't have it. And we have a robolab producing data. 24/7 - also one of the biggest repositories of data. And the most valuable data is the data from 40+ programs connecting Insilico, invitro, preclinical and clinical experiments. But even though we are probably the king of data, I still believe that the algorithm + ability to test quickly and in the right models is the main reason for our success to date.

My entire view of AI in drug discovery (for now) is this: we already have sufficiently capable algo-s to improve drug discovery. But... ... we do not have enough domain specific data to reach that. So, the race now is not for novel super AI, but for data acquisition capabilities (via building internal infrastructure for doing so, partnering, etc). It is costly. Hence, huge AI in DD company valuations. The 2024 and 2025 AI in drug discovery race is basically not AI race, it is data race. To be the first "OpenAI" of biotech. No single company on the market is still there.... but some are closer than others. If you want to know which are closer, stay tuned for the upcoming report next week (may be later, but hopefully it goes out before JP Morgan Health anyway). Image credit: myself (created it before gen AI era if anything...)

Re-growing a missing tooth? There is hope it will be possible one day 🦷. Next monday I am going to have one of my wisdom teeth pulled out 😱. So, I quickly searched the landscape and it appears there is a quite groundbreaking clinical trial in Japan probing a new drug candidate to stimulate the growth of a "third set" of teeth in humans, offering a natural alternative to dentures and implants. Humans, like other mammals, typically grow two sets of teeth. However, latent buds for a third set exist in our gums, arguably (I did not know that). The new experimental drug blocks USAG-1 protein aiming to activate tooth regeneration, as was successfully demonstrated in mice and ferrets. Some background info: Trials started at Kyoto University Hospital, with initial testing focused on adults for safety evaluation. The first clinical trial is targeting patients with congenital conditions, specifically those missing six or more permanent teeth from birth. This rare condition affects about 0.1% of the population. Importantly, the current results are only preclinical. So, there is no guarantee it would work in humans. But researchers aim to make this treatment accessible by 2030, particularly for children with specific conditions, if trials go well. If trials succeed, this could address tooth loss due to decay, disease, or injury! This research, led by Dr. Katsu Takahashi and his team give hope to me. Maybe, in future, we won't have easier ways to deal with dental issues... 🙏 (news in the comments)

Pretty sad, but AI-driven company Valo Health annouced it shelves diabetic retinopathy drug candidate after phase 2 failure. Valo Health announced it is halting the development of its ROCK 1/2 inhibitor, OPL-0401, after a Phase 2 trial in diabetic retinopathy failed to meet its primary endpoint. This marks a strategic pivot for the company, which will now focus on leveraging its AI-powered Opal platform. On the surface, the news is puzzling. After all, Valo's Opal is an AI-powered computational platform that integrates real-world patient data, human tissue models, and machine learning algorithms to identify and validate therapeutic targets and optimize small molecule drug candidates. At least from what I found online, it seems like one of the most sophisticated AI platforms out there... So, what happened? Here is possible reason. The failure of the candidate, OPL-0401, despite Valo Health's powerful AI platform, likely stems from its origin as a legacy, in-licensed asset that was not designed or optimized using Valo’s Opal computational framework. OPL-0401 was in-licensed from Sanofi in 2021, meaning it was developed using Sanofi’s drug development approaches (we don't know about them). This means the drug’s development may have lacked the Opal platform’s data-driven insights into patient stratification, target validation, and dose optimization—critical factors for diseases like diabetic retinopathy, which involve heterogeneous pathways and patient responses. Additionally, AI platforms like Opal excel in generating novel hypotheses and refining candidate molecules, but their full potential cannot be retroactively applied to pre-existing molecules without detailed, compatible datasets and integration. “While the data were intriguing, OPL-0401 did not incorporate elements of our platform in discovery or development,” this is what was said by Brian Alexander, M.D., the CEO of Valo Health in one of the articles. So, IMO, the failure in this case has business and strategy reasons, rather than tells much about the AI platform per se... but who knows? Let me know your thoughts... Image credit: Valo Health

Wrapping up 2024! 🔬 Gameto achieves the first live birth using Fertilo, its tech for maturing eggs outside the body, cutting hormone use by 80% and IVF cycles to 3 days; U.S. trials are underway. 💰 SandboxAQ raises $300M at a $5.3B valuation to advance AI-driven Large Quantitative Models for biopharma, finance, and cybersecurity, with recent collaborations and acquisitions. 🔬 Sanofi launches its 2025 iNext Awards in North America, offering up to $500K for 18-month immunoscience projects; submissions close January 30, 2025. 🔬 Tissue Dynamics ends 2024 with the largest human organoid screen, testing 7,680 liver organoids with 126 FDA-approved drugs and generating 20M+ data points. 🔬 NOETIK launches AI model OCTO-VirtualCell and visualization tool Celleporter to simulate cell behavior in tissues, aiding precision oncology and spatial biology. 💰 Ottimo Pharma, led by David Epstein, raises $140M in Series A to develop jankistomig, a PD1-VEGFR2 antibody, with IND filing planned for late 2025. 🔬 insitro earns $25M from Bristol Myers Squibb for identifying an ALS target using its AI-driven platform; total payments could exceed $2B as drug development progresses. 🚀 General Proximity, led by Armand Cognetta, raises $8M in seed funding to develop its proximity therapeutics platform for targeting undruggable proteins in cancer and aging-related diseases. 🔬 Gilead Sciences partners with Terray Therapeutics to use its AI-based tNova platform for small molecule drug discovery, combining Terray’s experimentation with Gilead’s expertise. 💰 Irish AI startup Nuritas raises $42M in Series C to expand its peptide discovery platform, Magnifier, which powers products like PeptiStrong and PeptiYouth. 🔬 Sanofi and Teva Pharmaceuticals report positive Phase 2b results for TL1A antibody duvakitug, showing remission in 47.8% of patients with ulcerative colitis and Crohn’s; Phase 3 trials planned. 🚀 ExpressionEdits, an AI-driven protein biotech, raises $13M and appoints ex CEO of Wave Life Sciences, Paul Bolno, as Board Chair to advance its Genetic Syntax Engine for gene therapies. 🔬 Nikhil Pradhan, a lawyer at Foley & Lardner LLP, outlines legal and governance strategies for AI drug discovery, focusing on IP, data protection, and regulatory alignment. 📈 An FTI Consulting survey shows 79% confidence in healthcare and life sciences for 2025, driven by M&A, GLP-1 treatments, and AI; cybersecurity remains a challenge. 💰 Syncromune® raises $100M in Series A to advance its immunotherapy platform for metastatic tumors, with Phase 1 trials underway. 🔬 WuXi Biologics resumes building its $300M Worcester plant after Congress failed to pass the Biosecurity Act targeting Chinese biotech. 🔬 SpiroChem CEO Thomas Fessard discusses how CROs adapt to market needs, focusing on bRo5 molecules, PROTACs, ADCs, and radiopharmaceuticals. Read more: https://lnkd.in/dR4Evuqg

Where do we stand in the artificial intelligence (AI) for drug discovery space?

I’m thrilled to announce our investment in Proton Intelligence a company that aims to transform the management of cardio-kidney-metabolic diseases! 🩺💰 🚀 The company is pioneering the first of its kind Continuous Potassium Monitoring (CKM™) platform, focused on patients with severe kidney impairment, including CKD and ESRD, and aiming to address a critical unmet need in managing cardio-kidney-metabolic diseases. LongeVC is proud to co-invest alongside SOSV We Venture Capital, Tenmile 15th Rock, Exor Ventures and Trampoline Venture Partners in a $6.95 million seed financing round. Quick insights: ✔️ The purpose of funding is to support ongoing clinical feasibility trials and prepare for a pivotal study in 2026. ✔️ Potassium fluctuations are life-threatening, contributing to therapy interruptions and higher mortality for CKD and heart failure patients. ✔️ CKD impacts 14% of U.S. adults, accounting for >25% of Medicare spending, with ESRD patients representing 7% of annual Medicare expenditures. We are excited to partner with Proton Intelligence in their mission to bring this innovative solution to market, because it may improve healthcare outcomes and reduce costs for millions of patients globally. Kudos to Sahan Ranamukhaarachchi , CEO of Proton Intelligence, and the team! Happy Holidays! 🎄 #Healthcare #MedTech Image credit: Proton Intelligence

Our first paper with Dominika Wilczok is now out in the journal of the American Society of Clinical Pharmacology and Therapeutics . This paper looks at the many inefficiencies in the AI-powered drug discovery industry and explains why we have not yet seen the first AI-discovered drug reaching approval even in the scenarios where companies cut corners by in-licensing or repurposing existing drugs for going after old targets. It contains valuable advice for the AI drug discovery and pharma companies as well as to the regulators. In the new year, let's work together to accelerate this industry and make real change. Referenced the work by the St Gallen Consortium Alexander Schuhmacher and Oliver Gassmann, BCG Madura Jayatunga and Chris Meier, BioPharmaTrend.com, Andrii Buvailo, Ph.D. and other analysts. The pharmaceutical industry lacks benchmarks on the preclinical R&D side. Trillions of dollars spent over decades and AIDD companies, investors, and pharma companies don't know what to optimize for and how to evaluate the companies. There are very few studies looking deep into preclinical R&D, time, cost, POS, and, most importantly, novelty, and potential patient and market impact. Without real benchmarks we will have many naked emperors walking around and since the program timelines often span decades and multiple carriers, no one will be held accountable and responsible.

In today's news, insitro, annouced a novel ALS target discovered using their high content screening platform, powered by machine learning. Insitro has announced the discovery and selection of a first novel target for Amyotrophic Lateral Sclerosis (ALS) as part of its ongoing collaboration with Bristol Myers Squibb. This milestone has triggered a $25 million payment to Insitro for target selection and discovery achievements. The discovery was driven by Insitro's proprietary CellML platform, a machine learning-enabled system designed to decode complex disease biology. Key technical components include: 200+ ALS Cell Lines: Engineered and patient-derived cell lines to generate motor neurons for disease research. High-Content Imaging: Advanced imaging capabilities, designed to support machine learning, to detect and analyze disease mechanisms. ML-Enabled POSH Technology: Machine learning-driven platform to explore genetic space at scale and identify genes that modify ALS. To me, it seems like Insitro is growing as a "disease modeling as a service" rather than pipeline-driven company. Which is cool in its own way. There are are similar companies, like CytoReason, that I could think about. Anyway, quite a lot of news lately in the AI drug discoveyr space. I think the next Where Tech Meets Bio (Substack Newsletter) is going to be a meaty one! (next monday, Christmas edition).

NOETIK has introduced OCTO-VirtualCell, a next-generation AI foundation model that simulates virtual cell behaviors in diverse tissue contexts. Alongside Celleporter, their interactive visualization tool, researchers can now explore how cells respond to different environments and gain insights into disease progression. Full article, including CEO Ron Alfa, MD, PhD insights, NOETIK's beautiful technical report and links to the tools: https://lnkd.in/gXy8fA-F