Matt Petroski

Google releases Tx-LLM, A Large Language Model for Therapeutics drug discovery and development. Tx-LLM rised from Google's Med-PaLM 2, and built to analyze a variety of chemical or biologics candidate to facilitate multi-modality driven drug-discovery pipeline, and with trained capability of navigating around some of the known toxicity. Free full text link: https://lnkd.in/etacc_5d Abstract: Developing therapeutics is a lengthy and expensive process that requires the satisfaction of many different criteria, and AI models capable of expediting the process would be invaluable. However, the majority of current AI approaches address only a narrowly defined set of tasks, often circumscribed within a particular domain. To bridge this gap, we introduce Tx-LLM, a generalist large language model (LLM) fine-tuned from PaLM-2 which encodes knowledge about diverse therapeutic modalities. Tx-LLM is trained using a collection of 709 datasets that target 66 tasks spanning various stages of the drug discovery pipeline. Using a single set of weights, Tx-LLM simultaneously processes a wide variety of chemical or biological entities (small molecules, proteins, nucleic acids, cell lines, diseases) interleaved with free-text, allowing it to predict a broad range of associated properties, achieving competitive with state-of-the-art (SOTA) performance on 43 out of 66 tasks and exceeding SOTA on 22. Among these, Tx-LLM is particularly powerful and exceeds best-in-class performance on average for tasks combining molecular SMILES representations with text such as cell line names or disease names, likely due to context learned during pretraining. We observe evidence of positive transfer between tasks with diverse drug types (e.g., tasks involving small molecules and tasks involving proteins), and we study the impact of model size, domain finetuning, and prompting strategies on performance. We believe Tx-LLM represents an important step towards LLMs encoding biochemical knowledge and could have a future role as an end-to-end tool across the drug discovery development pipeline. Google DeepMind

18

Latest on MDMA rejection; Novo vs. Lilly earnings; Recursion, Exscientia to merge; Q&A with Genentech CEO; and more https://lnkd.in/gRuf3XRp 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 [...] #LosAngeles #OrangeCountyCA #Biotech #Lifescience #News

1

Pick of the week from J Med Chem: covalent Bfl-1 inhibitors from AstraZeneca (Simon Lucas) for oncology indications. A covalent strategy to maximize selectivity within protein family and to gain potency in a more efficient way (lower MW, better LipE) than previous PPI inhibitors cases. The structure-based optimization effort was fueled by findings in MS and X-Ray screenings to deliver cell-active compounds #medchem https://lnkd.in/eeudWpC9

30