QTL Search

For a model with $k$ QTL and information criterion $I(L_k,k,n)$, we search for $k+1$st QTL over all intervals that do not presently have a QTL in them. For each of these intervals, we walk along the interval and calculate the information criterion $I(L_{k+1},k+1,n)$ for the presence of a QTL. We keep track of the minimum information criterion (equivalent to the maximum likelihood) within each interval. When all intervals have been tested, the minimum over intervals is determined and compared to the information criterion of the $k$ QTL model. If $I(L_k,k,n) - I(L_{k+1},k+1,n)$ is greater than the threshold, the QTL at that site is retained in the model. The process repeats until no new QTL are retained.

If MImapqtl is run in scan mode (S in position 0 of the work code), then the quantity $I(L_k,k,n) - I(L_{k+1},k+1,n)$ is written to the output file for each tested position. This yields a likelihood ratio profile giving the strength of evidence for a new QTL at the tested sites.

For crosses with three marker genotypic classes, both an additive and a dominant effect are added when testing for a new QTL. One can scan for additive effects only with the A option used in position five of the work code. If a D is used in this position, the search for dominance effects is restricted to previously identified QTL.