GenEpi: gene-based epistasis discovery using machine learning

Abstract

Background: Genome-wide association studies (GWAS) provide a powerful means to identify associations between genetic variants and phenotypes. However, GWAS techniques for detecting epistasis, the interactions between genetic variants associated with phenotypes, are still limited. We believe that developing an efficient and effective GWAS method to detect epistasis will be a key for discovering sophisticated pathogenesis, which is especially important for complex diseases such as Alzheimer's disease (AD). Results: In this regard, this study presents GenEpi, a computational package to uncover epistasis associated with phenotypes by the proposed machine learning approach. GenEpi identifies both within-gene and cross-gene epistasis through a two-stage modeling workflow. In both stages, GenEpi adopts two-element combinatorial encoding when producing features and constructs the prediction models by L1-regularized regression with stability selection. The simulated data showed that GenEpi outperforms other widely-used methods on detecting the ground-truth epistasis. As real data is concerned, this study uses AD as an example to reveal the capability of GenEpi in finding disease-related variants and variant interactions that show both biological meanings and predictive power. Conclusions: The results on simulation data and AD demonstrated that GenEpi has the ability to detect the epistasis associated with phenotypes effectively and efficiently. The released package can be generalized to largely facilitate the studies of many complex diseases in the near future.

Machine learning for the life-time risk prediction of Alzheimer’s disease: a systematic review

“… 41 Values across all studies were assessed to be below this threshold. The study with the highest EPV of 9.43 was Chang et al 42 The lowest EPV, 0.0018, was found for Wei et al 29 …”

Imputation and Reanalysis of ExomeChip Data Identifies Novel, Conditional and Joint Genetic Effects on Parkinson’s Disease Risk

“…To identify joint effects of SNPs on PD risk, we used a recently developed computational package called GenEpi [ 3 ], which applies a gene-based machine learning approach to discover pair-wise epistasis associated with a phenotype. In GenEpi, the first step is to group genetic variants by a set of loci (i.e., genes) in the genome using gene information available in the UCSC human genome annotation database [ 14 ] followed by dimensionality reduction of genetic features in each locus using LD which involves grouping of features into LD blocks using a given r 2 and D ’ threshold and selection of the features with the largest MAF to represent each block.…”

“…The genetic resolution of those large genotyped datasets can be increased via imputation of unobserved common and rare variants with advanced genotype imputation software which use new dense whole genome sequence (WGS)-based haplotype reference panels [ 2 ]. In addition, new methodologies have been recently developed using machine learning approaches to efficiently discover joint genetic effects of variants contributing towards complex disease risk by tackling the challenges of traditional GWAS, i.e., low statistical power to detect conditional and interaction effects due to the high dimensionality and multiple-test burden [ 3 ].…”

“…We accessed a large powerful PD NeuroX dataset available in dbGaP (dbGaP Study Accession: phs000918.v1.p1) that comprises 5540 cases and 5862 controls of European ancestry and first applied careful quality control procedures. Then, the dense WGS-based Haplotype Reference Consortium (HRC) reference panel was used to perform genotype imputation followed by standard GWAS association analysis, conditional analysis and interaction analysis using PLINK and R. Lastly, the recently published python package—GenEpi: a machine learning approach for gene-based epistasis discovery [ 3 ]—was utilised to identify joint genetic effects associated with PD.…”

Gene action, genetic variation, and GWAS: A user-friendly web tool

“…Epistatic interactions have been proposed as a potential source of genetic variation [18,19]. This leads to the third misconception that higher-order gene-gene interactions, which could, in principle, be detected by machine learning algorithms [20,21], would explain a lot of genetic variation. However, data and theory show that it is almost certainly not true and that additive variance is expected to remain the main contributor of genetic variation, even in the presence of high-order interactions [8,13].…”