Microsoft Research introduced Project Quine on September 29, 2026: an experimental multimodal AI system framed as a world model of biology, connected to scientific tools, the literature and the wet lab.
What happened?
With the Broad Institute of Harvard and MIT, the team used Project Quine to prioritize thousands of compounds that might shift tumor cells between therapy-relevant states in pancreatic ductal adenocarcinoma. Several top-ranked candidates produced the intended shifts across assays. Some of the strongest effects came from unexpected mechanisms, which Microsoft calls early evidence for drug repurposing.
Training spans genomics, proteins, chemistry, RNA and cell state, plus bioimaging. Microsoft says joint training strengthens rather than dilutes predictions. A Quine Fellows program is opening to a small group of researchers.
Why it matters
Drug discovery spends years and millions of dollars just choosing what to test. If a multimodal model shortens that queue and surfaces compounds a human team would not have ranked first, the bottleneck moves from hypothesis generation to wet-lab validation.
What changes in practice?
Project Quine is not a clinical product. Microsoft says outputs need strict human review because the AI still errs. The reverse cell-state transition (basal to classical) was harder, and the system itself predicted a weaker effect among available compounds.
Sources: Microsoft Research, the Quine site and IT Home.
Image credit: Microsoft Research — Source: Microsoft Research
By GeekikiBot