Microsoft Research unveils Quine, an early-stage AI system integrating multimodal biological data to prioritize lab experiments

Microsoft Research introduced Quine, an early-stage research effort to build a multimodal world model of biology that connects insights across biological scales and modalities. The system is intended to help scientists computationally search large hypothesis spaces and prioritize candidates for lab validation.

In collaboration with researchers at the Broad Institute of Harvard and MIT, Quine was used to prioritize compounds predicted to drive therapeutic tumor-state shifts, with several top-ranked candidates validated across multiple wet-lab assays.

Quine consists of two components: a world model that learns shared representations across biological modalities such as sequence, structure, function, cellular state, and imaging data, and an interactive harness that connects the model to orchestration and reasoning models, scientific tools, the literature, and researchers.

The system reflects the view that biology cannot be neatly divided into isolated domains, and that multimodal training strengthens generalization and supports broader biological reasoning tasks by enabling evidence from one modality to inform predictions in another.

Quine is explicitly positioned as experimental research technology, not intended for clinical or medical use, and its outputs require review and validation by qualified researchers.

Microsoft Research also announced the Quine Fellows program, which will provide a cohort of scientists access to the system to accelerate their research and provide feedback as the technology matures.