HIGH-RESOLUTION OF QUANTITATIVE TRAITS INTO MULTIPLE LOCI VIA INTERVAL MAPPING

A very general method is described for multiple linear regression of a quantitative phenotype on genotype [putative quantitative trait loci (QTLs) and markers] in segregating generations obtained from line cros ses. The method exploits two features, (a) the use of additional paren tal and F-1 data, which fixes the joint QTL effects and the environmen tal error, and (b) the use of markers as cofactors, which reduces the genetic background noise. As a result, a significant increase of QTL d etection power is achieved in comparison with conventional QTL mapping . The core of the method is the completion of any missing genotypic (Q TL and marker) observations, which is embedded in a general and simple expectation maximization (EM) algorithm to obtain maximum likelihood estimates of the model parameters. The method is described in detail f or the analysis of an F-2 generation. Because of the generality of the approach, it is easily applicable to other generations, such as backc ross progenies and recombinant inbred lines. An example is presented i n which multiple QTLs for plant height in tomato are mapped in an F-2 progeny, using additional data from the parents and their F-1 progeny.