Kalin Vetsigian

The paper introduces PanExonNet, a deep learning framework that predicts splicing patterns by integrating sequence and splicing factor expression, showing improved context specificity and generalization to new cell types.

The paper develops a deep learning approach to identify simple causal relationships in gene expression by finding low-dimensional representations that predict genetic perturbation responses.

The paper quantifies reproductive success variability in bacterial populations and finds heavy-tailed offspring distributions due to stochastic dormancy exits.