Authors: R Bernardo
Publish Date: 2012/12/28
Volume: 126, Issue: 4, Pages: 999-1009
Abstract
In composite interval mapping of quantitative trait loci QTL subsets of background markers are used to account for the effects of QTL outside the marker interval being tested Here I propose a QTL mapping approach called G model that utilizes genomewide markers as cofactors The G model involves backward elimination on a given chromosome after correcting for genomewide marker effects calculated under a random effects model at all the other chromosomes I simulated a trait controlled by 15 or 30 QTL mapping populations of N = 96 192 and 384 recombinant inbreds and N M = 192 and 384 evenly spaced markers In the C model which utilized subsets of background markers the number of QTL detected and the number of false positives depended on the number of cofactors used with five cofactors being too few with N = 384 and 20–40 cofactors being too many with N = 96 A window size of 0 cM for excluding cofactors maintained the number of true QTL detected while decreasing the number of false positives The number of true QTL detected was generally higher with the G model than with the C model and the G model led to good control of the type I error rate in simulations where the null hypothesis of no marker–QTL linkage was true Overall the results indicated that the G model is useful in QTL mapping because it is less subjective and has equal if not better performance when compared with the traditional approach of using subsets of markers to account for background QTL
Keywords: