One key goal of our group is to allow a broad research community to easily generate testable hypotheses and obtain biological insights from high-throughput genomic data. Over the last several years, we have developed several popular bioinformatics tools, including (i) TCPA: an integrative bioinformatic data portal for visualizing and analyzing cancer functional proteomic data (Li et al., Nature Methods, 2013); (ii) SurvNet: a valuable web tool for identifying network-based biomarkers that most correlate with patient survival data (Li et al., Nucleic Acids Research, 2012); (iii) BM-Map: a Bayesian stochastic model for accurately mapping RNA-seq reads and the user-friendly related software package (Ji et al., Biometrics, 2011); and (iv) PATHOME: an algorithm for accurately detecting differentially expressed subpathways (Nam et al., Oncogene, 2014).

Our department is one TCGA Genomic Data Analysis Center (GDAC). We have pioneered a series of pan-cancer comparative analyses using TCGA data. (i) We systematically assessed the prognostic and predictive utility of TCGA genomic and proteomic data (Yuan et al. Nature Biotechnology, 2014). This study provides deep insights into informative molecular data for building better prognostic models and selecting effective targeted therapy. (ii) We identified the system-level common properties of prognostic genes in gene co-expression network analysis across cancer types (Yang et al., Nature Communications, 2014). (iii) We analyzed the expression profiles of transcribed pseudogenes and assessed their biomedical relevance in multiple cancer types, highlighting pseudogene expression analysis as a new paradigm for investigating cancer mechanisms and discovering prognostic biomarkers (Han et al. Nature Communications, 2014). (iv) We performed the first comprehensive comparative analysis of somatic copy number alterations (SCNAs) across cancer types and revealed two distinct classes of SCNA breakpoint hotspots driven by different evolutionary mechanisms (Li et al., Human Molecular Genetics, 2012). These studies gain substantial insights into the common and distinct characters of human cancer.

RNA editing is an important post-transcriptional mechanism that can cause “mutations” at the RNA level, but which is largely ignored in the field of cancer research. We first develop a computational pipeline to systematically detect RNA editing events from TCGA sequencing data. We then systematically characterize key RNA editing events using experimental approaches. Our results lay a critical foundation for developing and implementing a novel class of RNA-editing–related biomarkers and therapeutic targets in cancer.

Next-generation sequencing (NGS) data has revolutionized modern biomedical research. We aim to identify driver molecualr events for tumor growth and progression from NGS data. We performed the first whole-transcriptome sequencing on gastric tumors: through a multi-layer and integrative analytic approach, we identified the loss of AMPKα2 as a driver event in the development of gastric cancer, and experimentally demonstrated its translational relevance as a potential therapeutic target for early-stage gastric cancer in Asian patients (Kim et al., Cancer Research, 2012). In another study, we developed the first systems-biology approach to identify driver mutations in endometrial cancer by combining whole-exome sequencing, high-throughput cell viability assays and high-throughput protein expression arrays. This approach addresses a long-standing challenge in the field of cancer genomics (Liang et al., Genome Research, 2012).