Please join us for this week’s Biostatistics & Medical Informatics seminar on Friday!
Time: Friday, October 22 2021 at 12:00-1:00pm
Zoom Link: https://uwmadison.zoom.us/j/92178324978?pwd=aW00MnZyNWVFaXhkVDVXUUpIUVRFZz09
Title: Integrative machine learning for regulatory genetic variation
Presenter: Dr. Alexis Battle, Department of Biomedical Engineering and Computer Science, Johns Hopkins University
Abstract: Understanding the genetic basis of human disease will require understanding genetic effects first at a cellular level. The majority of disease-risk genetic variants are non-coding — in order to link them to target genes and pathways, we seek to identify which genes’ expression they disrupt and under what conditions. Population studies of gene expression have provided thousands of expression quantitative trait loci (eQTLs) where genetic variants are associated with expression of a target gene. While eQTLs serve as a valuable resource for investigating possible gene targets of disease loci, key obstacles remain. I will discuss two major challenges we seek to address in our lab, including novel methods development and data collection to address each. First, eQTL studies do not address rare genetic variants, thus excluding tens of thousands of variants per individual found as frequencies too low to test in association studies. Here, we develop methods for integration of functional data with personal genome sequencing to better characterize rare regulatory variants. Second, even among common variants, over half of identified disease loci do not coincide with any known eQTL, with efforts frustrated by the dynamic, context-specific, and cell-type dependent nature of gene regulation. Here, we investigate data and methods for temporal and single-cell transcriptomics on a population scale. Together, our work aims to improve understanding of regulatory genetic effects for both common and rare variants and their influence on mechanisms underlying disease.