Exploiting linkage disequilibrium and other auxiliary information to improve genotype calling. (360G-Wellcome-090058_Z_09_Z)

We intend to develop methods to call genotypes that use both allele signal intensity from arrays and linkage disequilibrium information from public reference data such as the HapMap (Frazer, Ballinger et al. 2007) and 1 000 Genomes (Siva 2008) projects. This will involve a combination of the haplotype phasing component of imputation techniques (Marchini, Howie et al. 2007; Howie, Donnelly et al. 2009; Marchini and Howie 2010) and the traditional cluster analyses involved in genotype calling methods such as CHIAMO. ·As an initial step, a framework similar to BEAGLECALL will be implemented using IMPUTE for the haplotype phasing combined with a modified version of CHIAMO. An additional source of information will come from individuals who have been sequenced in addition to having SNPs assayed on arrays, this is to likely occur with some individuals from the 1000 Genomes Project. Sequencing data will allow many of the assay SNPs to be known with a high level certainty; this information can then be used to calibrate genotype calling algorithms. A particular focus will be given to rare variants, where perhaps only several heterozygotes and no homozygotes are present for the minor allele. Current methods will typically discard these genotypes as missing due to there being insufficient observations to characterise the heterozygote cluster. Newer generations of SNP chips will assay a greater number of these rare variants and the accurate calling of these genotypes will be crucial in allowing researchers to investigate the relationship between rare variation and phenotype.