BioPharmaTrend.com

In today's Where Tech Meets Bio (Substack Newsletter), we examine AI’s transparency problem in drug discovery, McKinsey & Company's case for modernizing clinical trials, a breakthrough in AI-designed enzyme engineering, and a new method for ultrafast protein mapping in tissues. 💠 Measuring AI’s Role in Drug Discovery – STAT News' Brittany Trang investigates how AI-driven biotech companies frame their capabilities, focusing on Absci and Generate:Biomedicines. The article explores the distinction between de novo AI-generated molecules and modified antibodies, emphasizing the need for clear benchmarks. Model Medicines' AIDD Code proposes six criteria, including identifying new targets, generating novel structures, and achieving high early testing success rates. The industry’s transparency gap remains a concern, with companies like Insilico Medicine publishing preclinical benchmarks while many remain vague on AI’s impact. 💠 Modernizing Clinical Trials – McKinsey & Company highlights the need for R&D IT upgrades to streamline clinical trials. Fragmented data and manual processes slow progress, but modernized systems could cut study start-up times by 15–20% and trial durations by 15–30%, unlocking up to $50 billion annually. Their four-layer framework—analytics, applications, data, and infrastructure—stresses interoperability. While some biopharma firms have begun IT modernization, substantial ROI remains a challenge. Success depends on technical upgrades and organizational shifts like staff training and workflow optimization. 💠 AI-Designed Enzymes Complete Full Catalytic Cycle – Institute for Protein Design, University of Washington, led by David Baker, used RFDiffusion to design serine hydrolases capable of completing full catalytic cycles—an unprecedented feat. Traditional static models fail to capture catalytic shifts, but by combining RFDiffusion with AlphaFold2 and PLACER, researchers improved functional design success rates from 1.6% to 18%. Though still slower than natural enzymes, this marks a step toward AI-generated biocatalysts. 💠 Mapping Millions of Cells in a Day – Massachusetts Institute of Technology's CuRVE (Controlled Uniform Rapid Visualization of Elements), detailed in Nature Biotechnology, enables ultrafast protein labeling in whole organs within 24 hours. Traditional methods are slow and uneven, but CuRVE, implemented in the eFLASH system, accelerates antibody movement via stochastic electrotransport and deoxycholic acid modulation. The study revealed discrepancies between antibody and genetic fluorescence-based labeling, suggesting the former provides a more accurate protein snapshot, with implications for neuroscience, pathology, and regenerative medicine. Full highlights: https://lnkd.in/d-W6Sy2S

In a recent study published in Nature, our scientists discovered that IL-27 — an important signaling protein that regulates immune responses — can boost the tumor-killing function of CD8+ T lymphocytes (CTLs), also known as “killer T cells,” without causing inflammatory side effects. Additionally, our scientists observed that IL-27 works cooperatively with blocking PD-L1 to stop cancer growth. Upon analyzing publicly available gene expression data, they further observed a link between IL-27 expression and better clinical outcomes in immunotherapy settings. These findings suggest that IL-27 plays an important supportive role for tumor-specific CTLs and offers a promising new avenue to strengthen current cancer immunotherapies. Learn more about their work: http://spr.ly/6044IdQkI

In today's Where Tech Meets Bio (Substack Newsletter) we explore the future of advanced therapeutics, the growing impact of combination oncology treatments, and a major Alzheimer’s prevention trial. 💠 The Future of Advanced Therapeutics – Moderna’s Sai Prathyusha Bhamidipati-MS, RAC-US, LSS-GB explores how mRNA vaccines, cell and gene therapies, and personalized medicine are reshaping treatment for complex diseases. While mRNA vaccines offer rapid adaptability, scaling and regulation remain hurdles. Meanwhile, over 2,000 gene therapies are in development, but viral vector production and safety concerns need ongoing attention. Personalized medicine promises better outcomes, but flexible manufacturing and regulatory shifts are key to success. 💠 Combination Therapies in Oncology – ICON plc’s Andreas Dreps and Bea Mann analyze how multi-drug regimens are transforming cancer care. With over 60 FDA oncology approvals in 2024, new strategies include immune system activation (e.g., ICIs with CAR-T and mRNA vaccines) and multi-targeted attacks (e.g., adagrasib + cetuximab for KRAS-mutant colorectal cancer). Key challenges include complex trial designs, dose optimization, and biomarker integration. 💠 Alzheimer’s Prevention Antibody Trial – Washington University in St. Louis launches a global study testing Eli Lilly and Company’s remternetug in high-risk young adults, aiming to stop #Alzheimer’s before symptoms appear. The trial, backed by $130M, will assess whether early amyloid removal prevents disease progression, with participants as young as 18. Full highlights: https://lnkd.in/dEz7rgWF

In today's Where Tech Meets Bio (Substack Newsletter): 💠 From LinkedIn to Drug Discovery: Reid Hoffman and Dr. Siddhartha Mukherjee launch Manas AI, an AI-driven biotech startup backed by $24.6 million in seed funding. Targeting aggressive cancers, Manas is designed as an end-to-end platform to identify novel targets and design first-in-class molecules through integration of AI and wet-lab biology. 💠 RNA-Based Biomarker Algorithms: Debiopharm and Genialis expand their collaboration to develop RNA biomarkers for WEE1-targeted cancer therapies, using Genialis’ molecular foundation model to analyze RNA sequencing data for improved patient stratification. In parallel, Deep Genomics employs its foundation model to explore RNA disease mechanisms for drug discovery and find candidates. In our recent interview, Brendan Frey, founder and CIO of Deep Genomics, discussed the challenges and opportunities of applying foundation models to RNA biology. 💠 Recursion Expands Into Clinical Tech: Recursion introduces its ‘ClinTech’ initiative to optimize clinical trials through real-world data and causal AI. In an interview featured in Genetic Engineering & Biotechnology News, Recursion’s Chief R&D Officer and CCO, Najat Khan, PhD, details how the company is applying this approach to several oncology programs, including REC-3565 (a Phase I MALT1 inhibitor), REC-1245 (a Phase I RBM39 degrader), REC-617 (a Phase I/II CDK7 inhibitor), and the upcoming REC-4539 (an LSD1 inhibitor). Notably, in academia, EPFL researchers led by Charlotte Bunne are also developing AI tools to improve diagnostic precision and treatment selection. Full highlights: https://lnkd.in/deZTf62u

In today's Where Tech Meets Bio (Substack Newsletter): 💠 Market Myopia in Oncology: Ryan Fukushima, COO at Tempus AI and CEO at Pathos, critiques the reliance on market forecasts in oncology drug development. He highlights how Imatinib, a breakthrough CML treatment, defied initial skepticism about its market potential. Fukushima argues for prioritizing biological insights over flawed market models. 💠 AI vs. Antimicrobial Resistance: A new report by the Fleming Initiative and Google DeepMind outlines how AI can tackle antimicrobial resistance (AMR), a crisis projected to cause 40M deaths by 2050. Key areas include rapid diagnostics, resistance prediction, and drug discovery. However, the report stresses the need for global collaboration, infrastructure investment, and interdisciplinary expertise to unlock AI’s potential. 💠 Virtual Tissues for Precision Medicine: Researchers at EPFL and ETH Zürich, including Charlotte Bunne, Johann Wenckstern, Eeshaan Jain, Gabriele Gut, Andreas Wicki, Kiril Vasilev, and Matteo Pariset, introduce VirTues, a foundation model framework for analyzing spatial proteomics data. VirTues leverages a vision transformer to create virtual tissue representations, supporting clinical diagnostics and biomedical discovery. Full highlights: https://lnkd.in/d745iWFu

IMO, 2025 will be a year of big data deals in pharma. For a starter, Truveta raised $320M to sequence 10M exomes and build largest genetic database 📀 🧬 As reported by Roman Kasianov for BioPharmaTrend.com, the health data company Truveta, in collaboration with Regeneron and Illumina, and 17 major U.S. health systems, has launched the Truveta Genome Project. The initiative aims to sequence the exomes of up to 10 million consenting participants, creating a diverse database that links genetic information with de-identified medical records. While the UK Biobank currently holds the world’s largest whole-genome sequencing dataset with 500,000 participants, the Truveta Genome Project focuses on exome sequencing—the protein-coding regions of the genome—at a much larger scale. The focus here is on diversity and representation. Thereby, the project aims to uncover insights that will support drug discovery, optimize clinical trials, and advance personalized healthcare. According to Aris Baras, senior vice president of Regeneron, with nearly three million exomes sequenced at the Regeneron Genetics Center to date, Regeneron scientists have already "identified dozens of genetic-based drug targets for conditions including chronic liver disease, obesity, cancer, and neurodegenerative diseases—many of which have progressed to clinical-stage treatments". The project will use biospecimens left over from routine medical tests, with Regeneron’s Genetics Center conducting the sequencing. Microsoft’s Azure platform will provide the cloud infrastructure to store and analyze the data securely. Interestingly, a key component of the project is the Truveta Language Model, an AI system designed to process and standardize large volumes of genetic and clinical data, built on Microsoft's Azure. In Truveta's Series C funding round, Regeneron has invested $119.5 million in the project, with Illumina contributing $20 million as part of a $320 million funding round that includes investments from 17 health systems such as Advocate Health, CommonSpirit Health, and Northwell Health. Truveta, founded in 2020, has already compiled de-identified electronic health records from 120 million patients through its partnerships with 30 health systems. Image credit: Truveta Read the news via link in the comments 👇

In the first newsletter of 2025: 💠 FDA's AI Draft Guidance: The FDA issues its first draft guidance for AI in drug development, proposing a 7-step framework for model evaluation and lifecycle maintenance. It highlights transparency, reliability, and early engagement with the FDA, reflecting insights from over 500 AI-related reviews since 2016. Public comments are open for 90 days. 💠 China’s Molecules: Eric Dai cites a Stifel Bank report by Tim Opler and team, revealing that one-third of big pharma’s molecules now come from China, up from nearly zero four years ago. 💠 2025 M&A Trends: EY’s Firepower report suggests life sciences M&A is ready for a rebound. While 2024 saw a shift to smaller, smarter, and 'more agile' acquisitions, big pharma retains $1.3T in deal capital. Patent cliffs and policy changes could drive larger moves, with AI partnerships and Chinese biotech collaborations gaining traction. Full stories here: https://lnkd.in/ddrnyb6H

In today's Where Tech Meets Bio (Substack Newsletter): 💠 Gameto’s Fertilo Milestone: Gameto, founded by Martin Varsavsky and led by CEO Dina Radenkovic Turner, reports the world’s first live birth using Fertilo, its ovarian support cell technology for ex vivo egg maturation. The process cuts hormone use by 80% and shortens IVF cycles to 3 days. 💠 Legal and Governance in AI Drug Discovery: As the FDA readies draft guidance for AI in drug development, Nikhil Pradhan of Foley & Lardner LLP outlines strategies for IP protection, data governance, and operational integration, based on Chris Bradbury's four-wave AI framework. 💠 NOETIK’s Virtual Cells: Noetik unveils OCTO-VirtualCell and Celleporter, tools for simulating cell behavior in tissues to study disease biology. Using AI, these models generate insights for precision oncology and beyond. Full stories here: https://lnkd.in/dR4Evuqg See you in 2025!🎄

Despite significant investment in AI-driven drug discovery and development, no AI-discovered drugs have yet reached clinical approval. A perspective by Alex Zhavoronkov (Insilico Medicine) and Dominika Wilczok published in the journal of the American Society for Clinical Pharmacology & Therapeutics (Clinical Pharmacology & Therapeutics), identifies 11 key challenges hindering progress, including the lack of preclinical R&D benchmarks, insufficient validation of AI models, and the imbalance between novelty and feasibility in drug development strategies. The authors propose solutions such as transparent benchmarking, end-to-end integration of AI with traditional methods, and prioritizing experimental validation. This work builds on analytical contributions by Andrii Buvailo, Ph.D. (BioPharmaTrend.com), Alexander Schuhmacher, Chris Meier, Oliver Gassmann, and Madura Jayatunga Explore a brief overview and the full paper in our newsletter: https://lnkd.in/dPYpxB5v

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.