Ordering dominant markers in F2 populations

Ordering dominant markers in F2 populations is considered a difficult problem. The difficulties arise from the fact that recombinations cannot be observed directly from the data. In general, the multi-point maximum likelihood would be the appropriate criterion for ordering markers. This criterion takes into account all available information present in marker data. However, calculation of multi-point maximum likelihoods is very time-demanding, especially if the number of markers is large. In this paper, ordering markers by minimising the number of recombinations between adjacent markers is used as a simple alternative to multi-point maximum likelihood. Contrary to multi-point maximum likelihood, this method does not involve any probability assumptions about the occurrence of recombination events. Simulated data indicate that the minimum number of recombinations between adjacent markers is approximately a linear function of the map length obtained by multi-point maximum likelihood. As a consequence, it will lead to more or less the same optimum marker orders. Optimisation of marker orders with regard to the number of recombinations between adjacent markers is carried out by a modified form of simulated annealing. The reliability of the resulting marker orders is studied by generating marker orders that are plausible with the data using a Metropolis algorithm