The Caveats

That's wonderful, isn't it? We have to do a little more work than for a traditional quantitative genetic analysis, i.e., we have to do a bunch of molecular genotyping in addition to all of the measurements we'd do for a quantitative genetic experiment anyway, but we now know how how many genes are involved in the expression of our trait, where ther are in the genetic map, and what their additive and dominance effects are. We can even tell something about how alleles at the different loci interact with one another. What more could you ask for? Well, there are a few things about QTL analyses to keep in mind.

• As currently implemented, QTL mapping procedures assume that the distribution of trait values around the genotype mean is normal, with the same variance for all QTL genotypes.¹¹

• QTL mapping programs often estimate the effects of each locus individually. It's not at all easy to search simultaneously for the joint effects of two QTL loci, although it's not too hard to look at the combined effects of QTL loci first identified individually. Composite interval mapping, in which additional markers are included as cofactors in the analysis, partially addresses this limitation. Multiple interval mapping looks at several QTLs simultaneously and shows some promise, but as you may be able to imagine it's pretty hard to search for more than a few QTLs simultanously.

• If some loci in the ``high'' line have ``low'' effects and vice versa, the effects of both loci (and possibly other loci) may be masked.

• Using this approach we can identify the QTL's that are important in a particular cross, but different crosses can identify different QTL's. Even the same cross may reveal different QTL's if the measurements are done in different environments. Methods to analyzes several progeny sets simultaneously are only now being developed.