These SNPs (pronounced snips) are locations in the human DNA sequence where two possible bases occur in the population. SNPs account for the most common type of variation in DNA sequence in humans and due to the recently developed high-throughput genotyping technology, genotype information on an individual's SNPs can be collected very cheaply.
Enter computational biologists around the world who have been devising ways to infer or extrapolate these haplotypes from the flood of genotype data produced by DNA sequencing efforts. Eskin and Ph.D. candidates Noah Zaitlen and Hyun Min Kang at UCSD, and research scientist Eran Halperin at ICSI, worked with NCBI scientists Michael Feolo and Stephen Sherry to infer haplotypes based on all of the data from genotyping studies deposited in NCBI's dbSNP database. Rather than use standard methods for inferring haplotypes, the computer scientists used HAP, a software tool originally developed at ICSI by Halperin and Richard Karp in collaboration with Eskin.
They ran the HAP algorithm on all dbSNP data sets using a cluster of 30 Intel Xeon processors provided by Calit2's National Science Foundation-funded OptIPuter project, in cooperation with the National Biomedical Computation Resource. Both organizations are based at UCSD. "In under 24 hours we were able to process more than 286 million haplotypes, partition those haplotypes into blocks, or regions, of limited diversity, and determine a set of 'tag' SNPs that capture the majority of genetic variation," explained Halperin.
The researchers' article appears in a special issue of Genome Research on "Human Genetic Variation," and its publication coincides with the release of a wide-ranging genotype study by the International HapMap Consortium in the journal Nature. The group's HapMap is a map of haplotype blocks and the ta
Source:University of Michigan Health System