Initiated an implementation of genetical genomics (expression quantitative trait loci eQTL) data analysis using microarray gene expression and genotype data. We developed an approach that uses a multivariate regression with the gene expression of known targets of transregulators (TRs) as response variables and individual single nucleotide polymorphisms (SNPs) as predictors to identify, compare and contrast TReQTLs in European (CEU), African (YRI) and Moroccan (MOR) populations. This work improves on the traditional approach when only the expression of one gene is associated with the genotypes of a SNP. The downstream target genes of the master regulator are typically co-expressed and share biological function. Therefore, it is practical to screen for eQTLs by identifying SNPs that are associated with the targets of a TR. This work also involved creating a database of the TR by SNP associations and building processing scripts to semi-automate the analysis pipeline. Lastly, we developed a novel way to infer the molecular pathways that regulate expression variants by ranking TReQTLs by the connectivity within the structure of Gene Ontology biological process subtrees that TReQTL SNPs, TRs and targets represent. -------------------------------------------------------------------------------------------------- Continued the development of procedures for the analysis of gene expression. A principal component analysis method for filtering Affymetix gene expression data was developed. Based on the amount of variation captured by the first principal component, probesets are removed from down stream analysis. This filtering step improves the false discovery rate as multiple testing for determination of differential gene expression is done on a substantially smaller gene set. -------------------------------------------------------------------------------------------------- Continued the bioinformatics collaborative support of investigators'research. 1) We devised of an analytical method to leverage multiple growth curve parameters obtained from yeast deletion mutants. The approach is based on biclustering whereby subsets of genes with highly similar growth curve outcomes are subset by a given perturbation (concentration of toxicant and exposure duration). In addition, we used methods to estimate the lag of growth and an approach to estimate the variance for a composite parameter. 2) We employed bioinformatics strategies to predict toxicity in the rat liver from exposure to acetaminophen and other hepatotoxicants using gene expression measures from the blood. 3) We developed a workflow for analysis and differentiation of gene expression data from mice exhibiting spontaneous hepatocellular carcinoma vs mice exposed to chemically-induced carcinoma. 4) We analyzed gene expression data from studies of gut microbiota in infants and yeast cells undergoing varying alterations in ribonucleotide incorporation.
Showing the most recent 10 out of 49 publications
