Presented at AACR SL in Montreal, Canada.
We are combining mechanistic modeling and machine learning to produce models that are both scalable and biologically meaningful.
Mechanistic models are explainable, contain biologically relevant end-points, and allow us to ask clinically-relevant questions, for example about penetrance and toxicity.
Machine learning models allow us to understand the effect of perturbations across on a much larger set of genes than with existing mechanistic models. They are currently also quicker to build.