QTL analysis: further uses of 'marker regression'.

A variety of approaches are available for identifying the location and effect of QTL in segregating populations using molecular markers. However, these have problems in distinguishing two linked QTL, particularly in relation to the size of the test statistic when many independent tests are performed. An empirical method for obtaining the distribution of the test statistic for specific datasets is described, and its power for demonstrating the inadequacy of a single-QTL model is explored through computer simulation. The method is an extension of the previously described technique of 'marker regression', and it is applied here to demonstrate two situations in which it may be useful. Firstly, we examine the power of the technique to distinguish two, linked QTL from one and compare this ability with that of two contemporary methods, 'Mapmaker/QTL' and 'regression mapping'. Secondly, we show how to combine information from two, or more, populations that may be segregating for different marker loci in a given linkage group. This is illustrated for two populations having in common just two linked marker loci although the sharing of loci is not a pre-requisite. Empirical tests are used to determine whether the same or different QTL are segregating and, if they are the same QTL, whether they are the same alleles. Evidence is discussed which suggests that the upper limit to the number of QTL that can be located for any single quantitative trait in a segregating populations is 12.