Multi-omics biomarker discovery platform
We have developed a multi-omics data and assay platform for rapid discovery of diagnostic biomarkers and drug targets. Our platform can process and extract features from a wide range of assay data, including genome, transcriptome, microbiome, immune repertoire, proteome and metabolome, from public and in-house generated data.
MarkerGenie is an online text-mining search engine that identifies
different types of biomarkers (including microbiomes, genes and metabolites) from the
publications in PubMed and PubMed Central.
We are developing algorithms and computational models to better characterize immune system and use the immune features to better predict disease state, drug response, adverse effect, drug resistance and to identify immunotherapy targets.
pMHC-TCR binding prediction
The recognization of the peptide-MHC (pMHC) by T-cell receptor (TCR) plays essential roles in adaptive immune response. Prediction of the interaction between pMHC and TCR can greatly facilitate the design of personalized T-cell therapy. However, due to the complexity of the problem and the scarcity of data, the prediction of pMHC-TCR binding remains challenging. We tackle this problem using deep-learning models trained on public and in-house generated data.
Neoantigen prediction is critical for personalized cancer vaccine development. Yet, its accurate prediction is not yet achieved. We are improving bioinformatic methods and prediction models, and meanwhile, developing assays to validate the findings.