Engineering cell state using artificial intelligence
Arc Institute
.
Abstract:
The virtual cell is a longstanding vision: a computational tool to guide experimental design and deepen our understanding of cellular function. In this primer, I describe the use of AI to engineer cell state, a capability essential for realizing this vision. Through a combination of foundation models built across biological scales and AI agents that guide data generation, I present a holistic platform for AI-guided biological design. I organize this primer around the three core axes of simulating cellular systems: representation, dynamics, and experimental agency. Drawing on my own contributions, I show how these advances enable us to search hypothesis spaces that have so far remained beyond experimental reach, with wide-ranging impact: from simulated whole-organism atlases of perturbed cells to patient stratification of drug response.
Biography:
Yusuf Roohani is Associate Director of Machine Learning at the Arc Institute. His research explores how artificial intelligence can guide experimental design in biological discovery, with a particular focus on predictive models of cell state. Yusuf earned his Ph.D. at Stanford University and prior to that, he spent four years at GSK as a Machine Learning Engineer working on early-stage drug discovery.