Epistatic effects refer to the interaction of allelic effects between different loci. Depending the natures of populations, the variance of epistatic effects for a quantitation trait may be partitioned into several different types of variance components, e.g., additive x additive, additive x dominance, dominance x additive, and dominance x dominance. Together with the main effect components, e.g, additive and dominance, there will be six genetic effect components in total.

The relative importance of each effect component usually varies across different traits. Accurate partitioning of the 6 genetic effect components can help us to understand the genetic mechanisms of complex traits and develop more efficient breeding programs.

Quantitative Trait Locus (QTL) mapping for epistatic effects has attracted much attention from geneticists. However, there is a great challenge in this field, i.e. the huge number of pairwise interaction effects to estimate, which result in a high computation burden. For this reason, current GWAS and QTL analysis rarely incorporate epistatic effects. Suppose the number of SNP marker(also called SNP bin) across the genome is m, there will be 2m main genetic effect while there will be 2m(m-1) epistatic genetic effects, which is beyond the computational capability for the current available QTL mapping programs, even only considering several thousand SNP markers.

Dr. Shizhong Xu at University of California, Riverside developed a new mixed-mode for mapping QTL with high accuracy by incorporating multiple polygenic covariance structure, which can be directly applied to polygenic-effect-adjusted GWAS in humans, plants and other species. Their mixed model require to calculate 6 kinship matrix including Ka, Kd, Kaa, Kad, Kda, and Kdd at first, then perform multiple polygenic component analysis for one target quantitative trait, and finally polygenic QTL mapping across genotypic markers. The polygenic QTL mapping include one dimensional Likelihood Ratio Test (1D LRT) estimation for Main effect QTL mapping and two dimensional Likelihood Ratio Test (2D LRT) estimation for Epistatic effect QTL Mapping.

Jointly working with Dr. Xu lab, we developed PEPIS (Pipeline for EPIStatic analysis) for polygenic, specifically epistatic QTL mapping. The PEPIS was implemented in C/C++ and employing parallel strategies for its intense computation modules. The full pipeline include two key sub-pipelines: sub-pipeline 1 for kinship matrix calculation and sub-pipeline 2 for Polygenic QTL Mapping.

sub-Pipeline 1 include three sequential steps: Step 1.1 for uploading the coded additive genotypic file usually named as Z.txt/Z.csv; Step 1.2 for uploading the coded dominance genotypic file usually named as W.txt/w.csv and Step 1.3 for Parameter configurations.

Notes: The genotypic file(s) are text file(s) and delimited by comma. All of the genotypic information in the files must be strictly stored as a format of m rows/lines and n columns (absolute no head or other supporting information), where m and n corresponds to the number of SNP marker/bin, and population individual respectively.

Sub-Pipeline 1 computes and output six text files corresponding to the six semi-diagonal kinship matrix.

Sub-Pipeline 2 include three sequential steps: Step 2.1 Uploading the quantitative phenotypic file that related to one specific traits; Step 2.2 Polygenic component estimation, which calculates and outputs the 6 polygenic components effect as vector to the target trait; Step 2.3 Main effect QTL mapping, which output a 1D LRT for genotype/SNP markers, and Step 2.4 Epistatic effect QTL mapping, which calculates and output a 2D LRT for genotype/SNP marker pair.

The final output LRT text files can be used to generate a LRT profile, then the genome-wide genetic information for one specific trait can be clearly displayed, which can in turn help breeder to understand the genetic mechanisms of complex traits and develop efficient breeding programs.

Notes: The phenotypic file is one line text file with n column and delimited by comma, where n corresponds to the number of population individuals.

Reference:

1. Xu, S., Mapping Quantitative Trait Loci by Controlling Polygenic Background Effects. Genetics, 2013. 195(4):p.1709-23.

2. Zhang W, Dai X, Wang Q, Xu S, Zhao PX, PEPIS: A Pipeline for Estimating Epistatic Effects in Quantitative Trait Locus Mapping and Genome-Wide Association Studies, 2016. PLoS Comput Biol, 12(5).