Skip to content
See the World Through ScienceA project of ALLATRA

Microsoft Opens a Biology AI That Learns Genes, Proteins and Cells Together

AI & Technology

Republish this story

Our work is licensed under Creative Commons BY-NC 4.0. You may republish this piece for free — with credit to ALLATRA Media and a link to the original, unedited beyond length trims, and not for commercial use.

Read the full license

Fluorescence micrograph of a cluster of pancreatic cancer cells, nuclei stained blue and a cytoplasmic marker in magenta against a black background.
Cultured pancreatic tumor cells under a fluorescence microscope, nuclei in blue and a marker of cell state in magenta (illustrative). Shifting cells like these from one state to another is the test case Microsoft describes."MIF-Induced Mesenchymal Marker in Pancreatic Cancer" by National Institutes of Health (NIH), via flickr, BY-NC · CC-BY-NC-2.0

Microsoft Research introduced Quine, an experimental artificial intelligence system for biology, on Sept. 29, 2026, and opened applications for a first cohort of outside scientists to use it. What is new is the shape: one model trained jointly across genetic sequence, protein structure and function, cell state and microscope images, not a set of specialist models wired together.

In its announcement, the company says the system is meant to rank and narrow options before anyone spends time in the lab, with the model sitting inside a harness that connects it to scientific tools, the published literature and the researchers using it. Microsoft calls Quine research technology only, not for clinical or medical use, and says its outputs may be incomplete or inaccurate and need review by qualified researchers.

Microsoft's worked example is pancreatic ductal adenocarcinoma, the most common form of pancreatic cancer. Working with researchers at the Broad Institute of MIT and Harvard, the company says it used Quine to rank thousands of compounds by their predicted ability to push tumor cells from one state into another, and that in its own wet-lab assays the highest-ranked compounds produced the largest shifts in the intended direction. Microsoft puts the whole exercise, from a wide field to a short list for the bench, at one weekend. These are the company's own figures: no paper, preprint or benchmark was published alongside the announcement, and no outside group has tested any of it.

The company also reports a result it says it did not expect. Several compounds moved cells toward a third state rather than either of the two it was tracking. Microsoft says the model predicted that, the lab bore it out, and pancreatic tumor cells may occupy a wider range of states than a two-way split.

Access stays narrow for now. Microsoft says availability is limited to the Quine Fellows program, whose first round of applications is open, and to selected research collaborations. The company says it expects to widen access later through its Microsoft Discovery product.

Sources

Spot an error?

Spot an error?

Report an error

Spotted a mistake on this page? Tell us what's wrong and our editors will take a look.

What kind of problem?

Only if you'd like us to be able to follow up. We won't use it for anything else.

We correct mistakes openly. Select any text to flag it. Fixes are logged under our Corrections Policy.

Report an error

Reporting on

Microsoft Opens a Biology AI That Learns Genes, Proteins and Cells Together

What kind of problem?

Only if you'd like us to be able to follow up. We won't use it for anything else.

We read every report. Corrections are logged publicly.