Remo Rohs Research Group

University of Southern California

How do three-dimensional molecular shapes determine where proteins bind DNA? The Rohs Lab develops machine learning and AI methods to model protein–DNA binding specificity using DNA shape features. The group builds and maintains DNAproDB, a database and processing pipeline for automated analysis and visualization of protein–DNA complexes. Researchers apply geometric deep learning and graph neural networks to predict protein–nucleic acid binding from structural and sequence features. The lab integrates computational structural biology, experimental mapping, and predictive modeling to reveal readout mechanisms of protein, RNA, and drug interactions.