DIGITAL BIOLOGY
Features
Roche outlines autonomous AI lab plans at its pharma investor day
At its Pharma Day, Roche outlined an Autonomous Lab Accelerator and a strategy of AI independence in R&D. Its presentation reports that 40% of pipeline decisions from the fourth quarter of 2025 through the second quarter of 2026 had a tracked AI or computational contribution. Roche says TargetNexus, its AI agent for target assessment, is on track to be part of 80% of research portfolio decisions by the fourth quarter of 2026.
AbbVie, argenx, Lundbeck and Takeda join Ginkgo and Apheris antibody developability consortium
The consortium aims to build a standardized dataset of 10,000 antibodies, because existing developability models have been limited by small, fragmented datasets. Members contribute proprietary sequences, Ginkgo runs the wet-lab characterization, and Apheris' federated infrastructure lets members train models without exposing raw sequences to one another. The initial dataset is targeted for delivery to members by early 2027.
MindRank signs China commercialization deal with 3SBio subsidiaries for its Phase 3 oral GLP-1
The agreement gives 3SBio subsidiaries exclusive commercialization of MDR-001 in mainland China, Hong Kong and Macau after approval and launch. MindRank, which describes the drug as AI-designed, receives RMB 150 million upfront, is eligible for up to RMB 1.6 billion in further milestones, and remains marketing authorization holder. MindRank says the candidate went from initiation to Phase 3 in 4.5 years on $23 million of R&D spend.
Octant announces $34.8 million in new funding for OpenADMET open-science initiative
The funding comes from the OpenAI Foundation, the Gates Foundation and ARPA-H, which triggered Phase II of its award a full year ahead of schedule. It supports high-throughput data on membrane transporters, metabolic enzymes, hERG and aminergic GPCRs, plus structural work at UCSF. Datasets, benchmarks and models are to be made broadly accessible, with community blind challenges to test predictive models.
10x Genomics announces Sentira, a cloud analysis platform that begins with spatial data
10x says GPU acceleration, using NVIDIA computing and the rapids-singlecell library, cuts the analysis iteration cycle in Sentira Spatial from hours on CPU alternatives to minutes. The launch coincides with Atera beginning to ship to customers. Early Access is free for researchers using Xenium or Atera, and a paid commercial launch is expected in early 2027.