Search all library resources

Multi-scale data-driven analysis of sex differences in human disease [electronic resource]

There are clear physiological differences between men and women, including dimorphism in disease susceptibility and treatment response. These disparities often stem from biological differences between the sexes. Sex matters at every level of biology, including genes, proteins, pathways and tissues. Sex chromosomes are fundamental determinants of genetic makeup; sex hormones regulate the expression of thousands of genes; immune response pathways differ between the sexes; and clinical variables like heart rate and pain intensity are also divergent. With current technological advances in high-throughput measurement modalities, we can simultaneously probe every gene in the genome or record millions of clinical features in databases for research. However, investigators using large-scale data sources often ignore the question of sex differences. Many experimental studies use only male animals, and clinical trials often exclude one sex in an attempt to reduce variability in their results. This has led to many pharmaceuticals failing to receive approval by the Food and Drug Administration due to toxicity or lack of efficacy in the untested sex. Rather than completely ignoring sex or viewing it only as a variable to control for, we should make sex differences research a priority of every study. In particular, we should incorporate sex analysis into large-scale genomics, proteomics and clinical analyses. In addition, integrating different types of data will enable more powerful mechanistic studies. Investigating sex differences can lead to new insights about how disease operates differently in men and women. These insights will in turn lead us closer to the goal of personalized medicine for both men and women. In this thesis, I describe methods to address informatics challenges in performing sex-differentiated analysis of high-throughput datasets. In particular, I show that I have (1) developed a novel statistical method to systematically analyze genome-wide association study data for sex differences, (2) applied the method to discover and validate a novel sex difference in a top Crohn's disease risk gene, (3) developed methods to mine large electronic medical record databases to discover sex differences in clinical pain measures, and (4) modeled sex-specific gene-gene interactions to discover molecular sex differences in Alzheimer's disease. This dissertation contains a set of methods for (1) genomics and other data-driven researchers to discover sex differences in molecular and clinical measurements and (2) sex differences researchers to integrate large-scale data sources. Many of the methods I have developed are generalizable to any situation in disease or genomics research with a binary variable of interest.