UCLA

The development of gene expression profiling technology has enabled the high-throughput discovery of the genes and pathways that underlie disease pathophysiology and phenotype. This work analyzes microarray and RNA sequencing data to identify genes and functional pathways associated with human diseases. In the first part, gene expression profiles derived from pancreatic ductal adenocarcinoma tumors are correlated to patient disease free survival time in order to find genes that confer a protective advantage. Four genes found to be significantly correlated with disease free survival were validated in tissue using PCR. In the second part, publicly available gene expression profiles for 16 skin diseases were integrated to build a disease classifier as well as characterize genes, functions, and pathways associated with each condition. Since data was drawn from different laboratories and experiment batches, we used Frozen Robust MultiArray Average to normalize the data and identified disease specific gene signatures using a ranking algorithm. Finally, we integrated this skin database with public data on interferon-regulated gene programs to find a negative inverse correlation between Type I and Type II interferon. The final part of this work applies the principles of comparisons in multiple diseases to the problem of characterizing subtypes of one disease. mRNA-seq techniques were briefly explored to probe for genes which historically have been difficult to detect on microarray. We compared microarray gene expression profiles from four subtypes of leprosy--lepromatous leprosy (L-lep), tuberculoid leprosy (T-lep), reversal reaction, and erythema nodosum leprosum--to build a proportional median-random forest classifier and perform functional analyses, such as weighted gene correlation network analysis (WGCNA), to find genes and pathways associated with each leprosy subtype. Integrating our proportional median subtype signature for T-lep with the WGCNA module associated with T-lep, we identified MMP12 as a novel differentiator of T-lep from L-lep. This gene was verified in tissue sections of leprosy using immunohistochemistry. The use of high throughput gene expression profile analysis in these three projects demonstrates the versatility and utility of transcriptome analysis when applied to human disease systems.