Microsoft introduces Quine for biology

On September 29, Microsoft Research introduced Quine. The authors are Nicolo Fusi and Jonathan M. Carlson. It is a research effort to build a multimodal world model of biology and an interactive harness that connects models, scientific tools, literature, and researchers. By world model they mean a system that can represent the state of a biological system, predict how that state changes after an intervention, and reason several steps ahead about the consequences. It does not replace experiments. Computation is for exploring, proposing, ranking, and prioritizing paths before scarce lab time is spent.

The concrete case is pancreatic ductal adenocarcinoma, the most common pancreatic cancer. With the Broad Institute of Harvard and MIT, they used patient-derived ex vivo models to test an existing hypothesis: tumor behavior and drug response depend on transcriptional cell state as well as genetics. In PDAC, tumor cells can sit in different states tied to how they respond to treatment.

Quine ranked thousands of compounds by how likely they were to push tumor cells between relevant states. Wet-lab work looked at the shift from classical to basal states. The highest-ranked compounds produced the largest intended shifts across assays. Narrowing the search, picking a handful of candidates, and validating them in the lab took one weekend. They write that this could save months of experiments and significant cost. That is a possibility, not a saving already counted. Some of the stronger effects came from compounds with unexpected mechanisms, described as early evidence for repurposing.

The reverse shift, from basal back to classical, was harder. Quine predicted that available compounds would have this weaker effect, and the experiments agreed. The model also predicted that several compounds would consistently move cells toward a third phenotype, and the lab saw that. The authors say the cell-state landscape is richer than a single classical-basal axis. The experiments tested hypotheses and produced new ones. Later work will use newly integrated RNA data to represent that landscape and to give calibrated confidence for which hypotheses are worth a wet-lab test.

Quine is experimental research technology, for research only, not for clinical or medical use. Outputs may be incomplete or inaccurate and need review by qualified researchers and experimental checks. Applications for the first Quine Fellows cohort are open. Early access is limited to that program and selected collaborations. They expect wider access later through products such as Microsoft Discovery. This piece did not open that product page.