Managing Challenging and Unexpected Cluster Patterns When Genotyping Plants and Other Complex Genomes

Background: The Axiom® Genotyping Solution for custom design of high-density SNP microarrays has enabled large-scale high-throughput SNP genotyping for a diverse spectrum of non-human genomes. Processing SNPs on this scale requires automated workflows, however non-human genomes, especially plants, present challenges: including polyploid genome structures, and high levels of nucleotide and structural diversity between subpopulations of a species. Most genotyping methods are designed to cluster intensities into three genotypes, namely AA, AB, and BB. However genomic diversity between subpopulations sometimes causes individuals to form more than three intensity clusters for the two segregating alleles. Such complex multi-cluster patterns result in a higher error rate of automated assignments of genotypes to clusters.

Result: The Affymetrix AxiomGT1 algorithm is used for automated clustering and genotype assignment. As a post-processing step we have developed statistical methods to identify SNPs with unexpected, multiple cluster patterns. These methods calculate the variance of each cluster and perform tests for unimodality, and are applicable to both diploid and polyploid genomes.  SNPs exhibiting such patterns are classified appropriately.  Our modified workflow returns an automatic division of polymorphic, high resolution markers from those markers that can benefit from further examination based on knowledge of the genetics of the sample under study.