Constructing and Joining Maximum Likelihood Genetic Maps

We present CART HAGENE a new piece of software for genetic mapping. Compared to existing genetic mapping softwares, CART HAGENE has both strengths and weaknesses. The most important weakness is certainly its actual limitation to backcross-like pedigree. However, when CART HAGENE applies, it is able to order markers and produce maximum multipoint likelihood marker maps with a large number of markers (eg. > 30) and a large percentage of missing data (eg. > 50%). This is done using optimization techniques originating from the artificial intelligence and operations research communities, hybridized with the usual statistical optimization algorithm EM. When a large number of markers are present, CART HAGENE offers a service comparable to the compare exhaustive search command of MAPMAKER at reasonable cost. More interestingly, when several data sets on the same organism are available, each with a different set of informative markers, CART HAGENE allows to automatically build maximum multipoint likelihood joined maps. On this type of problems, CART HAGENE compares favorably to JOINMAP, a much more general package dedicated to this type of problem. CART HAGENE has been used to build the first joined maps for the wasp Trichogramma brassicae and has also been assessed on randomly generated problems.