Research

Our research interests include bioinformatics, machine learning, cancer genomics, and pharmacogenomics. Our goal is to systematically model genomics and pharmacogenomics to better understand cancer biology and improve cancer therapy. Immersed in the cross-disciplinary environment of UPMC Hillman Cancer Center and the University of Pittsburgh School of Medicine, we team up with clinical, translational, and basic cancer scientists to bridge cutting-edge computational algorithms to unmet needs in precision oncology.

Deep learning of pharmacogenomics for adult and pediatric cancers – implementation of virtual screens


We aim at overcoming the barrier of applying emerging deep learning technologies to genomics and pharmacogenomics profiles when few samples are available. The hypothesis is that sophisticated deep learning models capture critical genomics features of a cancer cell or tumor to accurately predict its response to chemical (anti-cancer drugs) and genetic perturbations (CRISPR gene knockouts). We have designed two sophisticated deep learning models designed for genomics data and systematic chemical and genetic screens of cancer cell lines (Science Advances 2021 and BMC Medical Genomics 2019). The models are capable of capturing genomics features from high-dimensional multi-omics profiles of cell lines to accurately predict the response to hundreds of anti-cancer drugs and genome-wide gene knockouts. The cell line-trained models were transferred to predict tumors and achieve biologically meaningful results that were validated by clinical records. These are, as far as we know, the first models to successfully use deep learning to elucidate high-dimensional genomics and pharmacogenomics data. They could facilitate the prioritization of existing drugs for individual tumors and the search of novel therapeutic targets.

Systematic integration of genetic and pharmacologic cancer dependency maps

The rapidly growing cancer dependency maps pave the way to precision oncology by identifying and targeting the “Achilles’ heel” of cancer. There is a pressing need for software that systematically links such genetic (gene knockouts) and pharmacologic dependencies (small compounds). Addressing the need, we developed a web-based R Shiny app that incorporates heterogeneous data from large-scale high-throughput CRISPR screens, pharmacologic screens, and molecular signatures library, jointly covering 17k genes, 20k drugs, and 1k cell lines. The major goal is to match gene knockouts and drug treatments that induce similar effects in cell viability and/or gene expression perturbation in order to address two fundamental questions:

  • Which drugs can be potential surrogates to the knockout of a gene?

  • Which genes are potential targets or mechanisms of action of a drug?


The app has complementary and interconnected modules that address various query scenarios to identify potential druggable genetic vulnerabilities and understand the mechanisms of action of a known or new drug. Our Shiny app enables easy and systematic navigation, visualization, and integration of the rapidly evolving genetic and pharmacologic dependency maps of cancer. The results were presented in ISMB 2022 and ICIBM 2022. 🌟The tool will be online soon!🌟

Building prognostic biomarkers and study treatment resistance using integrative clinical genomics – myeloid malignancies as an example

Myeloid malignancies are heterogeneous diseases of abnormal myeloid progenitor cells. Over the past 10 years, we have been collaborating with hematologists and oncologists to study acute (acute myeloid leukemia; AML) and chronic forms (myelodysplastic syndrome; MDS) of these malignancies. The hypothesis is that integrative bioinformatics analyses of high-dimensional genomics profiles and clinical records lead to discovery of novel prognostic markers and therapeutic targets. Specifically, by integrative analyses of mutations, gene expression, and miRNA expression profiles of de novo AML patients, we showcased strategies of biomarker implementations based on different modalities of cancer multi-omics: miRNAs, mRNAs, gene regulations, and pathways. We developed a 3-miRNA prognostication panel to predict responses to standard chemotherapy and overall survival; these were validated in external cohorts (Leukemia 2015). Furthermore, we proposed for the first time and verified a novel mechanism: NPM1 (a frequently mutated gene in AML) gene mutation modulates miRNA−gene regulation that is associated with prognosis (Leukemia 2016). At the pathway level, we further showed the crosstalk among the pathways of Myc, OXPHOS, mTOR, and stemness governs patients’ response to “7 + 3” induction chemotherapy (European Journal of Haematology 2019). Overall, our studies bring biological and clinical insights into prognostic markers and chemoresistance mechanisms.

Besides hematopoietic malignancies, we have established collaborations to study metastatic castration resistant prostate cancer (mCRPC), hepatocellular carcinoma (HCC), and pediatric cancers.