My research group develops statistical algorithms and software for analyzing data generated from high-throughput genomics experiments.
Our projects involve close collaboration with biologists and clinicians. These collaborations play a key role to inspire our ideas and methodological development. We are currently focusing on developing methods and software for studying genetic interactions and network analysis using gene expression data including single-cell RNA sequencing (scRNAseq) and spatial transcriptomic technologies (STx). Standard differential expression analyses focus on comparing mean expression via a one-gene-at-a-time strategy. Arguably, changes in genetic interactions, sometimes without alteration in mean expression, have been reported to play a more critical role in downstream cellular transitions. Differential co-expression addresses this issue by examining whether there are correlated changes of expression between a set of genes under various biological conditions. This coordinated expression change suggests evidence for possible co-regulation related to the biological condition in question.

Differential co-expression analysis for STx data

Our work in this area is still ongoing. 


Robust approach for modeling non-linear co-expression patterns over time (TIMECoExpress)

The example below presents non-linear co-expression patterns over pseudotime trajectory during pituitary development. Our work in this area is still ongoing. 



Differential co-expression analysis for
scRNAseq data
Bayesian
  1. Single-Cell RNAseq Differential COExpression (scDECO) analysis framework
    Biometrics paper (2022)  scDECO software R package Tutorial
 
Frequentist

   2. Fast, Adaptive, Scalable, Tool for differential Co-Expression analysis (FAST-CoExpress)

     Paper  Software

Differential co-expression analysis for bulk RNA-seq data
  1.   Flexible bivariate correlated count data regression for bulk RNA-seq data 
     Statistics in Medicine paper (2020)  software

    2.   
Genome-wide search algorithms for identifying dynamic gene co-expression via Bayesian variable selection          
   
     Statistics in Medicine paper (2023) 
software

Integrative analysis of multi-omics data from single-cell experiment     Integrative multi-omics data analysis has been increasingly popular with the recent advances in technologies.     Because each data type usually has a distinct marginal distribution, a joint study of correlation presents a statistical challenge. We proposed a flexible framework for integrative analysis.

  1. Flexible copula model for integrating correlated multi-omics data from single-cell experiments
    Biometrics paper(2022) software           
Differential co-expression analysis for microarray data
  1. Modeling liquid association using gene expression data generated from microarray experiments
  Biometrics paper Software package

     2. fastLiquid Association algorithm

  BMC Bioinformatics paper Software package 

     3. Meta-analytic framework for modeling gene co-expression dynamics
  
  SAGMG paper Bioinformatics paper
Network Analysis
  1. nPARS: A comprehensive search algorithm for constructing Bayesian networks using large-scale genomic data
     Data: SNP, Gene Expression, and Cytotoxicity Outcomes ( ldocauc for the area under the log-dose response curve of docetaxel and lfuauc for the area under the log-dose response curve of 5-FU)
Gene expression amd cytotoxicity outcomes  are standardized ([x-mean(x)/sd(x)]).
  Paper  Algorithm: README, R source Code
  
 2.
Using gene expression to improve the power of genome-wide association analysis
   Paper  R Source code ReadMe