My group performs research to understand, through multiscale modelling, how organismal metagenomes specify their behavior and characteristics in conjunction with their environments. We accomplish this by developing novel computational biology and bioinformatics algorithms for predicting protein and proteome structure, function, interaction, design and evolution. We apply these basic science techniques to important practical problems in medicine, genetic and genomic engineering and nanobiotechnology.
I received the 2010 National Institutes of Health (NIH) Director‘s Pioneer Award to develop the Computational Analysis of Novel Drug Opportunities (CANDO) platform (http://protinfo.org/cando/) to repurpose drugs approved for other indications in a shotgun manner. Our integrated informatics platform determines interactions between and among all drugs and all protein structures to create compound-proteome interaction signatures. The compound-proteome interaction signatures are weighted using pharmacological, physiological and chemoinformatics data and compared and analyzed to predict the likelihood of the corresponding compounds being efficacious for all indications simultaneously, in effect inferring homology of drug behavior at a proteomic level.
Using this approach, we have made predictions for all the indications that our library of drugs maps to, with benchmarking accuracies that are two orders of magnitude better than what is observed when using random controls. We have performed prospective in vitro validations of our predictions, demonstrating comparable or better inhibition than existing drugs approved for clinical use in indications such as dengue, dental caries, diabetes, hepatitis B, herpes, lupus, malaria and tuberculosis. Our approach may be generalized to compounds beyond those approved by the FDA, and it can as well consider mutations in protein structures to enable precision medicine. We have also applied our computational techniques to design peptides for vaccines, antibacterial activity and inorganic substrate adhesion and model the structures, functions and interactions of all tractable proteins encoded by several rice genomes.
A consistent theme in our research is the combination of in virtuale simulation and homology inference, followed by in vitro and in vivo verification and application, directed toward holistic multiscale modelling of complex biological systems.