Descriptions:
This episode of the Latent Space AI for Science podcast features Bo Wang (SVP and Head of Biomedical AI) and Ci Chu (SVP of AI-Enabled Discovery) from Xaira Therapeutics, discussing the company’s newly released X-Cell virtual cell foundation model. The conversation explores how Xaira is using AI to predict how human cells respond to genetic perturbations and drug molecules—an approach the guests argue could transform drug discovery from artisanal trial-and-error into a rigorous engineering discipline.
The X-Cell model’s key capability is integrating multiple genome-wide perturbation campaigns simultaneously. The hosts highlight a striking result: combining seven such campaigns in a single model produces predictions that visually align far more closely with ground-truth experimental data than linear baselines. The discussion covers the evolution from bulk RNA sequencing to single-cell and spatial transcriptomics, with Ci Chu explaining the technical progression using an accessible analogy—from making a tissue smoothie that loses cellular identity, to cell-by-cell analysis, to spatially resolved measurements that preserve positional context within tissue.
Bo Wang and Ci Chu also outline Xaira’s three-platform AI strategy: protein design (rooted in co-founder Dr. David Baker’s lab at the University of Washington), virtual cell modeling for predicting biology, and patient representation models for matching individuals to therapies. The episode is essential listening for anyone tracking the frontier of AI applications in life sciences and computational biology.
📺 Source: Latent Space · Published July 21, 2026
🏷️ Format: Podcast
