The final weeks of the summer saw the SNP Consortium sign on several new players as it pushed forward with its plan to publish a genome-wide SNP-based linkage map by the end of the calendar year.
In two separate deals, the consortium signed Celera Genomics and its sister company Applied Biosystems to validate SNPs selected from the consortiums database and enlisted Motorola to provide genotyping services.
But the massive organization whose members include the Wellcome Trust, 10 major pharmaceutical companies, five prominent genomic research centers, Motorola, IBM, and Amersham Pharmacia Biotech is entrusting a four-member team from the department of genetics at Rutgers University with the guts of the project. Tara Matises Laboratory of Computational Genetics at Rutgers has been charged with the task of analyzing the resulting data and building the map itself.
The SNP Consortium has already identified and mapped over 1.5 million SNPs (of an estimated 3 million total), but the initial linkage map will use only 2,000 selected from the consortiums database and verified by Celera using the Applied Biosystems TaqMan reagent system. Motorola will use its CodeLink Bioarray platform to provide hundreds of thousands of genotypes.
But responsibility for building the map itself, which will not only indicate where SNPs are located on the chromosome, but map their relative positions on the basis of inheritance, rests solely on the computational prowess of Matises lab. The workhorse of the project will be MultiMap, a program that automates the construction of linkage maps that Matise co-developed with Aravinda Chakravarti of the Johns Hopkins School of Medicine.
The program is built on CRI-MAP, developed by Phil Green of the University of Washington a statistical approach for estimating the recombination-based distance between markers. MultiMap automates an overall algorithm for CRI-MAP, acting as a shell program that makes hundreds and hundreds of calls to CRI-MAP in the course of building a map, Matise explained.
MultiMap is freely available (http://compgen.rutgers.edu/multimap/index.shtml) and has so far been used to construct maps for mouse, rat, dog, horse, and some plant genomes. While many non-SNP linkage maps already exist, Matise said this is the first time her lab is using SNPs to make a genome-wide linkage map.
Noting that the mapping project is kind of a daunting task, Matise said that she has two things in her favor: One is the automated nature of the MultiMap program. Basically once the data files are set up and the parameters are decided on, the program does most of the work. In addition, Matise said she has ready access to a 64-processor machine at the university, so since there are 23 chromosomes I can actually be building 23 maps at the same time. While MultiMap itself isnt a parallel program, you can map each chromosome independently so its a parallel process.
Despite the difficulty of the task at hand, Matise said she is confident the map will be ready to publish by the SNP Consortiums December deadline.