Experimental and Computational Methods for Scaling-up Transcriptome Analyses and Improving Disease Risk Predictions

Project Summary Understanding the complex regulatory landscape of the genome will uncover fundamental principles of disease risk and etiology. Transcriptomic studies disentangle the functional nature of the genome by revealing the effects of variants on gene expression, but the cost and invasiveness of RNA-sequencing imposes limitations on the continued expansion of these studies. The demand to use data from genomics studies in the clinic is rising, but we have yet to establish methods of synthesizing genomics data in a way that improves clinical care. The long term goal of our work is to investigate environmental and genetic determinants of disease and to develop clinically meaningful ways of stratifying individuals according to these biological factors. Our central hypotheses are 1.) developing cheaper, more accessible methods of RNA-sequencing will enable massive scaling of transcriptomic studies and facilitate subsequent discovery from these studies, and 2.) using transcriptomic data for clinical predictions will augment current measures of genetic prediction, will provide key biological insights into disease mechanisms, and will increase portability of genetic risk scores across populations. In aim 1, we propose that sampling saliva, hair follicles, buccal tissue, and urine will allow for increased enrollment in transcriptomic studies due to the decreased invasiveness of sample collection, and we also put forward a low-cost RNA-sequencing method to overcome current financial barriers to study expansion. Aim 2 investigates the expression profiles of these non-invasive tissues and validates their use in understanding the genetic regulatory architecture of the body. In aim 3, we will generate novel risk scores from genetically predicted gene expression and from measured gene expression. These scores will be compared to the current standard for genetic clinical prediction, polygenic risk scores, and we will assess the predictive utility of these scores in multiethnic cohorts. We will further analyze differences between genetically predicted and measured gene expression to elucidate genetic and environmental mechanisms of gene expression regulation. Completion of this research proposal will produce methods central to improving our understanding of human phenotypes and will introduce ways of interrogating transcriptomic data that will yield essential biological and clinical insights.