I read with great interest the Commentary1 comparing mutagenesis and the dissection of quantitative trait loci (QTL). It reminded me of past genetic controversies2,3,4,5,6,7. Although I agree that the QTL strategy warrants criticism, in part to motivate efforts to overcome its difficulties, a balanced overview of the two strategies would have been more helpful in advancing quantitative genetic research.

Expression of many common diseases (such as hypertension, diabetes, alcoholism and so on) is influenced by both 'invariant' genes and 'variant' genes. The ability of the QTL strategy to find variant genes is a major advantage that justifies its use. QTL strategies can use various tools including phenotypic selection8,9 to focus directly on sets of naturally occurring variant genes and their interactions.

Mutagenesis induces variants, which include both the invariant genes and variant genes of complex traits. Considering the function of a highly complex system, mutagenesis can create perturbations in a subsystem by induced-mutation, which occurs with an extremely low probability in the natural variation of that subsystem, but it can have a major effect on another subsystem, therefore affecting the expression of the complex phenotype assayed. Although such an induced variant may be interesting to study what is possible in a system, it seems irrelevant in understanding the pathways of the naturally occurring expression of the phenotype. An induced mutation may give rise to a simalcrum of the phenotype, but we cannot rely on mutagenesis to deliver the genetic variants underlying many common diseases.