How Does a Model Choose the Next Biology Experiment?

Fabian Theis says a model chooses the next biology experiment by inverting a forward model. LabCompass, an inverse generative model from his group, takes a forward world model and says what to do in order to reach a chosen state.

Mapping cells came first. Predicting the next state came next. Steering the experiment is the step after that, and it puts the scientific method back in the loop. Papers often stop at a model, a concept figure, and a little validation, then a new project starts. Stopping there is not continual learning. Molecule design and material design already run the loop. Big pharma has named the theme, but he does not see many cases of several loops that run an experiment, update a model, and go back.

Picking the optimal next question is not trivial once the model exists. The hard part of LabCompass was the work with experimental colleagues at the University of Cambridge on hematopoiesis. They ran three or four loops and improved the model. A prediction still has to convince experimental people to spend time on a follow-up, and those experiments often take a while. A molecule is an easier ask because the stakes are lower. A new cell with strange additions is harder, and the fallback he describes is gut feeling. Reinforcement learning inside the loop is the vision. He would call the practice closer to active learning for now.