Threadmapper

The threadmapper software package is available at the Trheadmapper studio. The web page is described in Cheema, Ellis & Dicks (2010) THREaD Mapper Studio: a novel, visual web serverfor the estimation of genetic linkage maps. Nucleic Acids Research, 2010, 1–6, doi:10.1093/nar/gkq430.

The basic idea behind this software is that any marker data can be represented as a matrix of all pairwise inter-marker similarities. From these a measure of distance can be generated. In genetic mapping populations the relationship between map distance and this set of similarities depends on several factors:
(1) Whether the marker scores accurately represent allelic states in segregants.
A lack of correspondence can be considered as an 'error' but this may have a biological rather than technical source. For example, DNA methylation can change the apparrent allele in some marker assays, or DNA sequence data may derive from (physically and genetically) close duplicates of a particular sequence. That is the assay may not really correspond to a single locus.
(2) The loci corresponding to the marker may not be on the same linkage group. In this case the similarity or difference is a consequence of marker type, population type and the chance distributaion of assortments inthe (finite) population.
(3) For markers on the same linkage group the number of recombination events that occurred between the two loci in the generation of the population.

These three factors contribute to the similarity or dissimilarity in marker score, but have different meanings for the interpretation of map distance.

A symmetrical matrix of all possible pairwise distances can be examined by a variety of multivariate statistical methods. The matrix has the property that the entries are not really independent variables. This is usually considered as a problem in multivariate analyses giving rise to the infamous 'horseshoe effect'.

Fortunately this has at least one useful property: If for example all pairwise distances between adjacent marks on a ruler are calculated and this matrix is analysed by PCA, then the first few components capture much of the variation and, if these are plotted as a scatter plot, the points fall on the horseshoe shaped curve, and they are in the order they were along the ruler. This is precisely the solution sought from an analysis of pairwise distances between markers on a genetic map. Interstingly, with real or simpulate data the points are not quite on a smooth curve, but nearly so. This is because populations are not infinitely large. Major deviations from a smooth (threaded) curve represent other types of difficulty withthe data set. The treadmapper tool thus aims to assist with the construction of genetic maps and with the identification of problematic markers, or unusual genetic phenomena.

See: Course on Monday 17th October.