Our research is at the junction of statistical and biological sciences. Our research interests lie in two interrelated directions:
1) developing statistical methods for understanding biological questions, especially those related to large-scale genomic data;
2) identifying, formulating, and resolving important, yet not previously addressed statistical questions arising from the frontiers of biology.
The specific topics we have examined include:
  • Statistics:
    • Control of false discovery rates in multiple testing and asymmetric errors in binary classification
    • Measures of association 
    • High-dimensional linear model inference and variable selection 
    • Bipartite network stochastic block model inference 
  • Bioinformatics / Statistical Genomics:
    • Statistical rigor in omics data analysis
    • Statistical method development for single-cell omics data
    • Statistical method development for bulk short-read RNA-seq data 
    • Using statistics to quantitate the Central Dogma, the fundamental principle of molecular biology
    • Comparative genomics: developing novel statistical methods to investigate conserved or divergent biological phenomena in different tissue and cell types across multiple species 
    • Identification of gene-gene and protein-DNA interactions using diverse genomic data 

Featured Research

Statistical and computational method development for single-cell RNA sequencing data
The introduction of single-cell RNA-sequencing (scRNA-seq) over the last seven years has further revolutionized biomedical sciences by revealing the genome-wide gene expression levels within individual cells, in contrast to the traditional RNA-seq, now called the bulk RNA-seq, that only provides average gene expression levels across a batch of cells. The scRNA-seq technologies enabled researchers to identify cell subtypes, to elucidate the key genes that characterize each cell, and to demonstrate that cells are more heterogeneous than previously thought. However, the...
Learn more
Statistical and computational method development for identification and quantification of RNA isoforms/transcripts from short-read RNA-seq data

The advancement of next-generation high-throughput sequencing technologies has been revolutionizing genomic studies in the last decade. In particular, the RNA sequencing (RNA-seq) technology, which has deep coverage and base level resolution, enables investigating human and other eukaryotic species’ transcriptomes (i.e., sets of RNA molecules in cells) with unprecedented detail and clarity. Unlike previous technologies such as microarrays, RNA-seq provides information on alternative splicing (i.e., how is the DNA of one gene possibly transcribed...

Learn more
Using statistics to quantitate the Central Dogma

We have investigated how to use statistics to quantitate the Central Dogma: DNA makes mRNA, and mRNA makes protein, one of the most fundamental principles in modern biological sciences. There are two key synthesis steps: transcription (information flow from DNA to mRNA) and translation (information flow from mRNA to protein). In addition, mRNA and protein molecules are degraded (i.e., destroyed), making four steps in total. In the past seven years, I collaborated with Dr. Mark Biggin, a biochemist at Lawrence Berkeley National Laboratory,...

Learn more
Statistical methods and analyses for integrative and comparative genomics

We have developed two new statistical metrics “TROM” (TRanscriptome Overlap Measure) and “EPOM” (EPigenome Overlap Measure) to evaluate the similarity of transcriptomes andepigenomes (i.e., sets of chemical compounds that are not part of the DNA sequence but are on or attached to DNA) within and across species. In work with the modENCODE consortium [1, 2, and 3], we used TROM to discover a previously unknown conservation between the developmental stages of D. melanogaster and C. elegans, two vastly different model organisms...

Learn more
Statistical methodology in association measures, variable selection, and classification evaluation metrics

We have worked on statistical methodology development in the following projects, which were motivated by prominent biomedical research questions. The methods we developed have great potential to become critical tools in computational biology and medical informatics.

First, we have developed a “new R²” association measure for describing complex gene interactions. Capturing important gene expression relationships is a key component in computational biology research (e.g. network study). Commonly used measures such as Pearson correlation and Spearman rank...

Learn more