One AI-guided round beat a seven-round directed-evolution benchmark. ProteinGuide - a method that helps existing AI protein-design tools generate better-targeted protein sequences. Read more: nature.com/articles/s41587-0…

Ardigen’s AI and computational biology capabilities will strengthen VERAXA’s in-house AI efforts to optimize cancer target pair selection for conditionally active T cell engagers and antibody-drug conjugates. Read more: ardigen.com/ardigen-veraxa-b… #AIforBiologics #DrugDiscovery

AWS launched an open-source tool that lets AI assistants like Claude Code or Kiro search this entire data catalog for you - just by asking in plain English. Read more: aws.amazon.com/blogs/opensou… Stay updated: ardigen.com/newsletter-signu…

Chan Zuckerberg Biohub has released an open-source atlas built with ESMFold2 AI model, covering 6.8 billion sequences and over 1 billion predicted structures - one of the largest computational resources for protein biology to date. Read more: nature.com/articles/d41586-0… #ainews

Generative models are becoming part of everyday pharma workflows. Bristol Myers Squibb has partnered with Anthropic to bring Claude to more than 30,000 employees worldwide. Read more: reuters.com/legal/litigation… Stay updated: ardigen.com/newsletter-signu… #biotech #AI #DataDrivenBiotech

One year after launching Elsa 1.0, the FDA has expanded its internal AI assistant organization-wide. Elsa 4.0 now supports: - Custom agentic AI - Data analysis & visualization - Document generation & text recognition - Secure web search capabilities fda.gov/news-events/press-an… #AI

AI is now guiding one of the most complex cardiac interventions - in real time. Philips has received FDA clearance for DeviceGuide, an AI tool designed to assist physicians during transcatheter heart valve implantation procedures. Read more: fiercebiotech.com/medtech/ph… #healthtech

A new AI approach (RBD-AIM) connects sequence → structure → epitope → function in one pipeline. Rather than focusing only on structure prediction, it allows prediction of antibody behavior, including binding properties and neutralization potential. nature.com/articles/s41467-0…

A team at Michigan State University has introduced GPS, a deep learning platform that predicts how small molecules affect gene expression, directly from chemical structure. Source: doi.org/10.1016/j.cell.2026.…

The upgraded Evo 2 has been trained on a massive atlas spanning all domains of life - from bacteria to humans. It doesn't just "read" code; it understands the structural "language" of life at a single-nucleotide resolution. Read more: nature.com/articles/s41586-0… #biotech #AI

AI is giving scientists a system-level view of the cell. Researchers from the Broad Institute of MIT and Harvard and ETH Zurich developed an AI framework that integrates multiple types of biological data to better understand how cells behave. Read more: news.mit.edu/2026/ai-help-re…

CardiOmicScore is a deep learning framework that integrates multiomics data to predict six major cardiovascular diseases up to 15 years before onset. Read more: nature.com/articles/s41467-0… #biotech #healthtech #AI #precisionmedicine #bioinformatics

Human - AI teams identified more cancer patients for clinical trials The study, published in Nature Communications, tested whether combining research staff with AI improves screening for oncology trials. Read more: nature.com/articles/s41467-0…

The Billion Cell Atlas is described as the largest dataset ever created to study how genes influence human cells. It aims to support drug target validation and provide large-scale experimental data for AI model training in drug discovery. Read more: firstwordpharma.com/story/70…

1m DNA bases. 6,000+ genomic predictions. One model. Google DeepMind just introduced AlphaGenome - an AI model that evaluates genetic variants across ~6,000 genomic signals, delivering results in under a second. Read more: nature.com/articles/s41586-0… bbc.com/news/articles/c39428… #AI

35 enzymes designed from scratch by AI. Researchers from TU Graz and the University of Graz report in Nature a new AI-based method for designing enzymes, without searching existing protein databases. Read more: technologynetworks.com/infor… Stay updated: ardigen.com/#form #AI

Eli Lilly is making its AI/ML platform, TuneLab, available through an integration with Benchling, giving more than 1,300 biotech companies access to models trained on over $1B of proprietary research data. Read more: prnewswire.com/news-releases…

AI-designed drug delivers late-stage clinical results Takeda reports positive Phase III results for zasocitinib, an AI-supported oral psoriasis drug targeting TYK2. A clear case of AI-assisted discovery translating into late-stage clinical efficacy Source: reuters.com/business/healthc…

$555M AlgenBrain-AstraZeneca deal AstraZeneca is licensing AlgenBrain, an AI platform that maps gene function to disease outcomes and helps surface targets for immune-related therapies. Read more: reuters.com/business/healthc… #AIinBiotech #GenomicMedicine #DrugDiscovery

Bringing AI supercomputing into drug discovery Lilly and NVIDIA are launching a DGX SuperPOD to train AI on millions of experiments and broaden therapeutic search. Through Lilly TuneLab, biotechs can use select models without sharing data. Read more: reuters.com/business/healthc…