Xia Ning

This paper introduces MHCrank, a deep learning model that improves MHC Class I antigen processing predictions using cleavage site-specific kernels and learned amino acid representations, outperforming existing methods.

The paper evaluates machine learning models for predicting proteoform retention and migration times in top-down mass spectrometry, finding high accuracy with Prosit and a neural network model.

The paper develops a joint model to recommend safe or to-avoid drugs in polypharmacy scenarios and creates datasets for evaluating drug-drug interactions.