My Research

My primary research interests are Machine Learning (ML) related problems such as Multiple-Instance Learning, AUC Optimizations, and Kernel Methods. I am interested in exploring potential applications of these ML methods in Bioinformatics and Immunoinformatics. My ongoing work aims at improving existing computational methods for accurately and efficiently predicting Kinase-specific Phosphorylations sites, Protein Subcellular Localization, Protein-RNA binding sites, and MHC-II binding peptides.

Motif based methods (i.e. PSSMs and HMMs) are widely used in bioinformatics tools. However, these methods can not model correlations among different positions in the biological sequence. In a recent study, we showed that ML algorithms outperform motif based methods when relatively enough training data is available. My future work aims at improving existing methods for modeling biological sequences.