Genome-wide association mapping including phenotypes from relatives without genotypes
MetadataShow full item record
A common problem for genome-wide association analysis (GWAS) is lack of power for detection of quantitative trait loci (QTLs) and low precision for fine mapping. Here, we present a statistical method, termed “ssGBLUP”, which increases both power and precision without increasing genotyping costs by taking advantage of phenotypes from other related and unrelated subjects. The procedure achieves these goals by blending traditional pedigree relationships with those derived from genetic markers, and by conversion of estimated breeding values (EBVs) to marker effects and weights. Efficiency of the method was first examined using simulations with 15,800 subjects, of which 1500 were genotyped. Comparison included two scenarios of ssGBLUP (S1 and S2), classical genome-wide association (CGWAS) and BayesB. For genomic evaluation, the highest accuracy of prediction was obtained by the second iteration of ssGBLUP. Power and precision for GWAS were evaluated by the correlation between true QTL effects and the sum of m adjacent single nucleotide polymorphism (SNP) effects. The best accuracy for QTL mapping occurred for ssGBLUP with m=8, and BayesB with m=16. For simulation data set, ssGBLUP is faster and easier for GWAS without computing pseudo data compared with CGWAS and BayesB. In the second and third studies, ssGBLUP was extended to GWAS on broiler chickens for single- and multi-trait model. Dataset consisted of 2 pure lines (L1 and L2) across 5 generations for 3 traits: body weight at 6 wk (BW6), ultrasound measurement of breast meat (BM), and leg score (LS) coded 1=no and 2=yes for leg defect. Single-trait model was only based on BW6 of L2. There were 294,632 and 274,776 individuals in pedigree for L1 and L2, of which 4667 and 4553 were genotyped using a SNP60k panel. Results of QTL mapping had express in format of Manhattan plots, which were constructed as proportion of genetic variance explained by each region consisting of 20 consecutive SNPs. Different peaks across traits and lines suggest different selection goals. The forth study analyzed distribution of differences between pedigree- and genomic-based relationship matrices (G-A). QC reduced differences and was able to identify parent-offspring conflicts. Large discrepancies between G and A imply unidentified errors or limited pedigree depth. From both simulation and application studies on GWAS, ssGBLUP approach is faster, simpler, and easily applicable to complex models including multi-trait, maternal effects, indirect genetic effects, and random regression.