Haplotype Heritability Mapping Method Uncovers Missing Heritability of Complex Traits

Masoud Shirali,Ricardo Pong-Wong,Chris S Haley,Pau Navarro,Sara A Knott

doi:10.1038/s41598-018-23307-4

Masoud Shirali, Ricardo Pong-Wong + Show 3 more

Open Access

https://doi.org/10.1038/s41598-018-23307-4

Copy DOI

Abstract

We propose a novel approach to analyze genomic data that incorporates haplotype information for detecting rare variants within a regional heritability mapping framework. The performance of our approach was tested in a simulation study based on human genotypes. The phenotypes were simulated by generating regional variance using either SNP(s) or haplotype(s). Regional genomic relationship matrices, constructed with either a SNP-based or a haplotype-based estimator, were employed to estimate the regional variance. The results from the study show that haplotype heritability mapping captures the regional effect, with its relative performance decreasing with increasing analysis window size. The SNP-based regional mapping approach often misses the effect of causal haplotype(s); however, it has a greater power to detect simulated SNP-based-variants. Heritability estimates suggest that the haplotype heritability mapping estimates the simulated regional heritability accurately for all phenotypes and analysis windows. However, the SNP-based analysis overestimates the regional heritability and performs less well than our haplotype-based approach for the simulated rare haplotype-based-variant. We conclude that haplotype heritability mapping is a useful tool to capture the effect of rare variants, and explain a proportion of the missing heritability.

Highlights

Analytical approaches using multiple SNPs jointly (e.g. SKAT9 or regional heritability mapping10) have been shown to have more power in detecting both common and rare variants than single SNP mapping methods[11]
All simulated architectures were analysed using either regional heritability mapping method (RHM) (where we used genotype information to construct the regional genomic relationship matrix (RGRM)) or Haplotype Heritability Mapping (HHM)
In all the simulation scenarios, the HHM approach provides more accurate estimation of regional heritability for all the genetic architectures simulated than the RHM approach (Fig. 3)

Summary

Introduction

Analytical approaches using multiple SNPs jointly (e.g. SKAT9 or regional heritability mapping10) have been shown to have more power in detecting both common and rare variants than single SNP mapping methods[11]. Multi-allelic markers such as haplotypes inferred from genotyped SNPs12,13, have been used to study the genetic structure of populations[14,15,16] and to detect un-genotyped causal variants[17,18,19]. Genotype imputation techniques have been presented to predict genotypes at variants that are not directly measured[20], they are dependent on a reference panel that has been densely genotyped to identify shared haplotypes between the reference and the target populations[21,22]

Methods

Results

Conclusion