Computational Inference of Genetic Epistasis Networks
Identifying causal single nucleotide polymorphisms (SNPs) contributing to phenotypic differences can dramatically improve our understanding of human disease susceptibility and etiology. However, causal epistatic SNPs whose functions need to be studied together, not in isolation, to devise effective therapeutic interventions. We will use statistical methods for mediation analysis to construct SNP interaction networks using public resources, such as Universal Protein Resource, Ingenuity Pathways Knowledge Base, Molecular Signature Database, and Kyoto Encyclopedia of Genes and Genomes. This network will provide the prior information needed for modern sparse machine learning models, such as Lasso, being developed by the ENIGMA Center. This collaborative project will leverage the strengths of the two centers with an innovative combination of knowledge of gene-gene interactions and sparse learning models for identifying causal SNPs.
News feature:
KnowEnG and ENIGMA centers partner to merge techniques from Big Data Powerhouses