The use of selective genotyping in the detection of quantitative trait loci (Q-TL) by sib pair analysis

This study covered two main areas in the methodology of Quantitative Trait Locus (QTL) detection with the use of genetic markers: i) the use of sib pair analysis for QTL detection in animal populations of large family size and ii) the use of selective genotyping in sib pair analysis for the reduction of genotyping costs. The regression of the squared phenotypic differences of sib pairs on the proportion of alleles they share Identical By Descent (IBD) at a marker locus can provide a test of linkage between a marker and a QTL. The t-values of the regression coefficients can provide a way to detect linkage. The regression coefficients can provide estimates of the QTL position and variance. The power of detection and parameter estimates (position and QTL variance) obtained from the analysis of progeny originating from populations with different family size, were compared using simulated data. Families of large size provide a large number of sib pairs, since the number of sib pairs is related to the family size. The correlation of phenotypic differences between sib pairs of the same family has little effect on the test statistic and consequently on the power of detection. In order to decrease the amount of genotyping without any negative effects on the power of detection, a different approach to sib pair analysis was used. Instead of using the associations between phenotypic differences and proportion of alleles shared IBD at a marker locus for the detection of linkage, any changes in the proportion of alleles shared IBD at a marker locus in a selected sample of phenotypically discordant sib pairs could be used to identify linkage between the marker and a QTL. This can be done by regressing the estimated proportion of alleles shared IBD at the marker locus minus the expected mean (0.5) on the phenotypic differences of the selected sample of sib-pairs minus the mean phenotypic differences of the all sib-pairs from which the sample has been drawn. The t-values of the regression constants were used to test any deviations of