Microsoft Research has announced Quine, an experimental research effort that aims to build a multimodal world model of biology together with an interactive harness that links models, scientific tools, literature, wet labs, and researchers. The system is intended to support iterative scientific workflows where computational predictions inform experimental design and experiments in turn improve the model.
What Quine is and how it works
Quine consists of two central components: a world model of biological systems and a harness that orchestrates reasoning models, scientific tools, and data sources. The design is intended to transform a scientist’s question into concrete proposals, turn proposals into experiments, and feed experimental measurements back into both the scientist’s next question and the model itself.
The world model learns shared representations across multiple biological modalities and scales, including sequence, structure, function, cellular state, and imaging data. By training jointly across these modalities, Quine can use evidence from one modality to inform predictions in another, avoiding siloed single‑domain specialists and improving generalization across a broader set of biological reasoning tasks. The system is explicitly presented as a tool to inform and prioritize experiments, not to replace wet‑lab work.
Application to pancreatic cancer cell‑state biology
A concrete application highlighted in the announcement involves pancreatic ductal adenocarcinoma (PDAC). In collaboration with researchers at the Broad Institute of MIT and Harvard, Microsoft Research used patient‑derived ex vivo models to test the longstanding hypothesis that tumor behavior and drug response depend on transcriptional cell state as well as genetics.
Using Quine, the team predicted and prioritized thousands of compounds according to their potential to shift tumor cells between therapeutically relevant transcriptional states. In wet‑lab assays that focused on the transition from a “classical” to a “basal” cell state, compounds ranked highest by Quine produced the largest intended shifts across multiple experimental assays. The entire pipeline — from rapidly narrowing the search space to selecting a handful of candidates for laboratory validation — was completed in a single weekend, potentially saving months of experimental time and substantial costs.
Some of the strongest effects were observed for compounds with unexpected mechanisms of action, indicating that the system may help uncover opportunities for drug repurposing and discovery. The reverse transition (basal to classical) produced weaker effects, which Quine had predicted. Experiments also revealed that several compounds consistently moved cells toward a distinct third phenotype, an outcome the model had indicated and that suggests the PDAC cell‑state landscape is more complex than a simple classical–basal axis. These experimental results both validated model hypotheses and generated new questions to incorporate into future model training.
Future work will integrate additional RNA datasets and tasks to better represent the richer landscape of cell states, improve state‑transition predictions, and provide calibrated confidence estimates to help researchers prioritize hypotheses for wet‑lab testing.
Access, the Quine Fellows program and responsible development
Quine is presented as experimental research technology for research use only, not for clinical or medical application. Microsoft Research cautions that Quine’s outputs may be incomplete or inaccurate and require review and experimental validation by qualified researchers.
Access to Quine will be phased and controlled. The announcement opens applications for an initial Quine Fellows program cohort to give selected scientists hands‑on access to the system and the opportunity to accelerate their own research while providing scientific feedback. Initial availability is limited to the Quine Fellows program and select research collaborations; Microsoft says it will apply internal reviews and built‑in safeguards as the system evolves. Over time, as the technology matures, Microsoft expects to expand access through products such as Microsoft Discovery.
Why this matters
Microsoft frames Quine as an attempt to change how biological discovery is practiced by enabling a continuously accelerating loop between models, experiments, and researchers. The company argues the real test of the system is how useful it is when confronted with real scientific practice — incomplete evidence, novel questions, and the need to prioritize limited laboratory resources. If effective, Quine could shorten the time from hypothesis to validated discovery while keeping scientists and experiments at the center of the process.
Quine builds on Microsoft Research’s long history of work at the intersection of computation and biology, spanning immunology, virology, genomics, biomedical imaging, cell biology, and protein engineering.
