DIGITAL BIOLOGY
Features
Anthropic previews a standard for AI agents to operate lab instruments
The Model Hardware Standard connects programmable devices such as microscopes, liquid handlers and robotic arms through a shared interface. Anthropic says it reduces integration work from weeks or months to hours or minutes and lets agents adjust parameters during experiments. Developed with HHMI Janelia Research Campus, it is initially available in research preview to scientific labs and advanced manufacturers.
BenchSci partners with Google Cloud to host its preclinical research platform
The multi-year agreement makes Google Cloud the primary infrastructure for EMET, BenchSci's agentic research environment. It brings a biological knowledge graph with 858 million nodes and 2.2 billion relationship edges into an environment with AlphaGenome and AlphaFold 3. The graph draws partly on licensed access to 16 million closed-access papers.
Transfyr launches with $25 million to capture laboratory work as machine-readable data
The seed round, led by General Catalyst, backs a company founded by Ginkgo Bioworks' former head of AI and ARPA-H's founding director. Transfyr aims to record the physical details and context of bench work that often go undocumented. It intends to turn those observations into data for AI systems and laboratory automation.
Adaptyv raises $40 million to expand automated protein testing
Highland Europe led the Series A for Adaptyv's laboratory platform, which tests proteins designed by AI models. The company reports fivefold growth in laboratory throughput and more than 100 customers, including pharmaceutical companies. The funding is intended to expand experimental capacity and help protein designers obtain the measurements needed to evaluate and improve their models.
AGENTEX robotically tests genetic codes without recoding an organism's genome
The system screens modified transfer RNAs in cell-free translation reactions, exploiting the ability of natural synthetases to charge many non-standard tRNAs. Researchers demonstrated polypeptide translation incorporating a non-standard amino acid and three reassigned codons. Their proposed 34-codon code could encode 20 canonical amino acids and leave 14 codons for reassignment; that expanded capacity remains a proposal.