EXPERIMENTAL PLAN

Principal Investigator/Program Director Williams, Robert W.

Data and map export

The NTB is intended to supplement rather than replace existing QTL mapping programs. In particular, it will make some existing programs more accessible to neuroscientists. Some existing programs, such as Mapmaker/QTL (Lincoln, Daly et al. 1992) and QTL Cartographer (Basten, Weir et al. 1994) , provide sophisticated methods, including maximum-likelihood evaluation and automated model-building.  However, they require the user to create specially formatted data files and marker maps. These requirements can be an insurmountable barrier for the nonspecialist. To enable neuroscientists to use these programs, the NTB will allow export of marker data, trait data, and marker maps in the formats required by these programs.

We will make public the genotype data generated by our Genotyping and Mouse Colony Core as soon as possible, in formats suitable for Map Manager QT and QTX, Mapmaker/QTL, and QTL Cartographer. Even before the proposed Web interface is in place, we will provide Map Manager QT data files for downloading from the <nervenet.org> site. Map Manager data files have been available for downloading from this site since early in 1994. Data sets include complete RI genotype files, complete MIT SSLP map files, and several other major mapping panels. A complete list of Map Manager files currently available online is at <www.nervenet.org/MMfiles/MMlist.html>.

Dominance models

Dominance refers to the fact that the trait value of a heterozygote may not be the mean of the values of the two homozygotes. This phenomenon can be analyzed in intercrosses, in which separate contributions from additive and dominance effects can be estimated. These effects are expressed in the trait equation as coefficients in two terms, one in which the three genotypes (maternal, heterozygous, and paternal) are coded as 1, 0, and 1, and one in which the genotypes are coded as 0, 1, and 0. For interval mapping, the expected genotypes are typically nonintegral and are derived from the genotypes and distances of flanking markers.

only one. The NTB will allow a choice of these models for the target QTL in simple or composite interval mapping. For the background loci in composite interval mapping, it will allow any of the constrained models, that is, those that use one degree of freedom.

Interference models

Genetic interference refers to the observation that, at least in some organisms, two recombination events close to each other are much less common than would be expected by chance. This fact affects the estimation of a QTL genotype from the genotypes of flanking markers. No completely satisfactory model of interference has been described, and even if one were available, its implementation would probably be computationally demanding. However, we can easily implement two extreme models, one in which interference is absent and one in which it is complete (multiple recombinants are absent, at least within an interval).

In recombinant inbred lines and advanced intercrosses, recombination events are derived from multiple rounds of mating, and since adjacent recombinations may be derived from different meioses, interference is assumed to be absent. In addition, the probability of recombination is greater than the nominal map distance between markers (Jiang and Zeng 1997) . QTL mapping in recombinant inbred lines and in the G₁₀ advanced intercross, therefore, will be analyzed by a no-interference model that accounts for the map expansion from multiple rounds of mating.

Fast interval mapping

Next Topic

• Weighted regression.