Patrick A. Murphy

This paper uses deep learning to quantify dynamic self-organization in Myxococcus xanthus by encoding microscopy images into interpretable features that predict developmental phenotypes.

This study challenges the traffic jam model of Myxococcus xanthus aggregation by showing streams are not essential for aggregate formation but may accelerate it, with a genetic component to streaming.

The paper analyzes 1D aggregation models of bacterial populations with non-exponential reversal times, showing they improve aggregate formation and persistence.

The paper develops a deep-learning model to convert phase-contrast images into fluorescence-like images for better analysis of bacterial aggregation and rippling behaviors in Myxococcus xanthus.