I am interested in developing and applying computational research methods in biology. In particular my focus has been on protein sequence and three-dimensional structure.
I have developed a template-free protein structure prediction program called SSThread. SSThread predicts the structure of contacting pairs of α-helices and β-strands and then assembles overlapping pair predictions to generate an ensemble of core structures. Loops and side chains conformations are then predicted.
I have also created DeepPPP (Deep learning for Protein Property Prediction) that predicts multiple protein structure properties. Predicted properties include 8-state secondary structure, half-sphere exposure, real value backbone torsion angles, contact maps and domain boundaries. Predictions are made by deep neural networks including embedding, convolutional and recurrent layers.