← Back to Explore

CisRF: A Novel Machine Learning Approach to Predict Context Dependent Impacts of Disease Associated Regulatory Elements on Gene Expression

JSHS · 2024

Overview

Genome Wide Association Studies have identified 560,000+ Single Nucleotide Polymorphisms (SNPs) associated with disease, most of which are not in protein -coding regions of the genome. However, the biological mechanism behind these non-coding genetic associations remains largely unknown. Uncovering these mechanisms requires understanding impacts of genetic variants on gene expression which are tissue and cell type dependent, but wet-lab experiments to measure these impacts are costly and time-consuming. To a ddress this problem, I developed a computational method, CisRF, to enable fast and low -cost evaluation of context -dependent effects of non -coding SNPs on gene expression across diverse tissues and cell types. CisRF models gene expression as a function of a ctivities of non -coding DNA regulatory elements via Random Forests (RF), trained using public ENCODE data with 414 pairs of RNA -seq and DNase-seq samples. Using principal components and cross -validation to adaptively optimize model complexity, CisRF achieved higher prediction accuracy compared to a naive RF in 73% of genes. Upon applying CisRF to hypertension and height-associated non-coding SNPs, I evaluated how perturbations of SNP-containing regulatory elements change gene expression in 97 distinct tissues and cell lines. CisRF correctly predicted that heart ventricle samples had significantly larger gene expression changes than other tissues in hypertension related SNPs (two -sample t -test, p -value=0.0038), while myocyte samples had larger gene expression changes for height -related SNPs (p -value=0.00069). This demonstrates CisRF’s ability to efficiently screen large numbers of contexts to identify potential mechanisms through which non - coding genetic variants influence phenotype, which was not possible with conventional wet -lab experiments.

Competition history

  • JSHS 2024 Category not listed

Resources

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Source: Junior Science and Humanities Symposium

Save projects to your library

Sign in with Google to keep track of projects you find interesting, organized into folders. An account also raises your daily allowance for “Has this been done?”, and lets you create a key for the MCP server with a much higher limit than anonymous use. Browsing stays public.

Continue with Google