Statistical Methods for VINOtyping on Affymetrix® Axiom® Arrays

Variable Intensity Non-hybridization Oligos or VINO’s (Didion et al.; BMC Genomics 2012, 13:34) is a class of hybridization probes, used in microarray-based genotyping, whose sequence is significantly dissimilar to the targeted genomic region of some species’ subpopulation(s). This dissimilarity between probe and target can thus lower hybridization intensities such that genotypes of the subpopulation are often incorrectly called as heterozygotes by genotyping algorithms that cluster intensities into three expected genotypes, namely AA,AB and BB. VINO probes tend to group members of the divergent subpopulation into a fourth cluster denoted as the V cluster. Therefore, correct identification of this V cluster, a process described as VINOtyping, is a critical step.

We have developed a statistical method called “VINOCaller” to VINOtype all VINO SNPs based on the posterior genotyping information generated by the automated clustering algorithm AxiomGT1 which is used by the Axiom^® Genotyping Console^™ Software. VINOCaller uses posterior genotypes as the prior information to initiate the Baum-Welch algorithm to iteratively search for the optimal locations of the four clusters AA, AB, BB and V. It then derives genotypes and VINOtypes when convergence is reached. We applied VINOCaller to genotype data from multiple species and demonstrate VINOCaller can successfully identify the V cluster in the presence of all three genotypes (AA, AB and BB) or in the presence of only two genotypes (AA and BB).  VINOCaller is available from Affymetrix in a SNP post-processing R package “SNPolisher” and is applicable to Axiom^® genotyping arrays.