DIFFERENCE BETWEEN SINGLE-STEP AND WEIGHTED SINGLE-STEP GBLUP IN GWAS FOR BODY WEIGHT OF RAINBOW TROUT Oncorhynchus mykiss

In genome wide association studies (GWAS), a group of individuals, with a given measured phenotype, is genotyped using a dense SNP panel in order to statistically verify the association between the markers and the trait of interest. Some statistical appr oaches are used to perform GWAS. For single-step GBLUP (ssGWAS ) and weighted single-step GBLUP (wssGWAS) approaches the phenotypes from non-genotyped individuals are considered in the analyzes,  but in wssGWAS, more than one interaction is performed using the SNP weights estimated in the previous analysis as start for next analysis. This methodology can avoid spurious solutions of SNPs, reducing the "noise" in the results and increasing the proportion of genetic variance explained by the markers. We performed a GWAS analyses using ssGWAS  and w3ssGWAS approaches aimed at identifying  genomic associated to body weight of rainbow trout.

Rainbow trout of 105 full-sib families from year class 2011 of Aguas Claras (Puerto Montt , Chile) breeding program were used. Body weight at 18 months (BW18M = 169.39 ± 36.19g) of 5,005 animals were recorded. G enotyping  of  4,596 animals was performed using an Affymetrix ® 57K SNP panel. The quality of the genotypes was filtred considering Hardy-Weinberg Equilibrium (p-value < 1×10-3), Minor Allele Frequency (MAF < 0.05) and genotyping call rate ( SNP  and samples)  < 0.90. GWAS analysis  was performed  with trhee interactions using BLUPF90 statistical software. For the first iteration, the weights for each SNP used as the start was 1 (ssGBLUP ), in the second iteration the weights estimated in the first iteration for each SNP were used as the start, and finally, for the third interaction the weights of each SNP used as start were those estimated in the second interaction (w3ssGWAS). To identify regions of the genome associated with the analyzed traits , the percentage of genetic variance explained by 20 adjacents SNPs was accumulated in windows (20 SNP window).

The ssGBLUP analysis identified  important genome regions for BW18M at chromosomes 21 and 7 where the top windows explained 0.8% and 0.53% of the genetic variance respectively. U sing w3ssGWAS approach, the top windows located chromosomes 24, 15 and 21 explained 3.02%, 2.74% and 2.52% of the genetic variance respectively (Figure 1).  In conclusion, the different statistical approaches resulted in changes at the genomic regions associated with BW18M and increase in the percentage of additive genetic variance explained by SNPs

Figure 1. Manhattan plot of genetic variance explained (%) by 20 adjacents SNP window for BW18M of  rainbow trout using ssGBLUP (left) and w3ssGWAS (right) approaches.