Localising Loci underlying Complex Trait Variation Using Regional Genomic Relationship Mapping

Abstract

The limited proportion of complex trait variance identified in genome-wide association studies may reflect the limited power of single SNP analyses to detect either rare causative alleles or those of small effect. Motivated by studies that demonstrate that loci contributing to trait variation may contain a number of different alleles, we have developed an analytical approach termed Regional Genomic Relationship Mapping that, like linkage-based family methods, integrates variance contributed by founder gametes within a pedigree. This approach takes advantage of very distant (and unrecorded) relationships, and this greatly increases the power of the method, compared with traditional pedigree-based linkage analyses. By integrating variance contributed by founder gametes in the population, our approach provides an estimate of the Regional Heritability attributable to a small genomic region (e.g. 100 SNP window covering ca. 1 Mb of DNA in a 300000 SNP GWAS) and has the power to detect regions containing multiple alleles that individually contribute too little variance to be detectable by GWAS as well as regions with single common GWAS-detectable SNPs. We use genome-wide SNP array data to obtain both a genome-wide relationship matrix and regional relationship (“identity by state" or IBS) matrices for sequential regions across the genome. We then estimate a heritability for each region sequentially in our genome-wide scan. We demonstrate by simulation and with real data that, when compared to traditional (“individual SNP") GWAS, our method uncovers new loci that explain additional trait variation. We analysed data from three Southern European populations and from Orkney for exemplar traits – serum uric acid concentration and height. We show that regional heritability estimates are correlated with results from genome-wide association analysis but can capture more of the genetic variance segregating in the population and identify additional trait loci.

Genome-wide association and regional heritability mapping to identify loci underlying variation in nematode resistance and body weight in Scottish Blackface lambs

“…On the other hand, several exclusive regions reaching the suggestive level were identified only with RHM. Nagamine et al (2012) showed that RHM performed better than a standard GWA study, especially when associated SNPs do not have large enough effect to be declared significant at the genome-wide level. In principle, estimating the trait heritability for chosen regions allows integration of the variance contributed by both rare and common variants into a single estimate of additive variance.…”

“…Nagamine et al (2012) suggested that the use of smaller windows can improve mapping resolution. In our study, this was not always the case; that is, when moving from 100 to 50 SNP wide windows, only in a few cases, there was an improvement in the level of significancesometimes regions became not significant.…”

“…Few have attempted to formally estimate the distribution of marker effects with dense SNP markers (for example, Hayes et al, 2010;Kemper et al, 2011) and thus the required power of experiments for complex disease traits, such as nematode resistance, is still unknown. An alternate approach exploiting dense SNP chip data, known as Regional Genomic Relationship Mapping or Regional Heritability Mapping (RHM) (Nagamine et al, 2012), has been advanced as a better approach to capture more of the underlying genetic effects. This method provides heritability estimates attributable to small genomic regions, and it has the power to detect regions containing multiple alleles that individually contribute too little variance to be detected by GWA studies.…”

The Nature of Genetic Variation for Complex Traits Revealed by GWAS and Regional Heritability Mapping Analyses

“…These include the sequence kernel association test (SKAT) (Wu et al 2011) and multiple-SNP methods such as regional heritability mapping (RHM), a variance-component approach based on restricted maximum-likelihood (REML) estimation of additive variance explained by genomic regions of a given size (Nagamine et al 2012). A comparison of the power of these two methods and other composite P-value methods and single-SNP methods has been made recently by Uemoto et al (2013).…”

Submitted summaries

“…Unfortunately, GWAS seems to localise a relatively small proportion of the total genetic variation in the traits of interest (e.g., Kemper et al, 2011). An alternate approach proposed by Nagamine et al (2011), known as Regional Genomic Relationship Mapping (RGRM), may better capture genetic effects. This method provides heritability estimates attributable to small genomic regions, and it has the power to detect regions containing multiple alleles that individually contribute too little variance to be detected by GWAS.…”