A draft fur seal genome provides insights into factors affecting SNP validation and how to mitigate them

Abstract

Custom genotyping arrays provide a flexible and accurate means of genotyping single nucleotide polymorphisms (SNPs) in a large number of individuals of essentially any organism. However, validation rates, defined as the proportion of putative SNPs that are verified to be polymorphic in a population, are often very low. A number of potential causes of assay failure have been identified, but none have been explored systematically. In particular, as SNPs are often developed from transcriptomes, parameters relating to the genomic context are rarely taken into account. Here, we assembled a draft Antarctic fur seal (Arctocephalus gazella) genome (assembly size: 2.41Gb; scaffold/contig N50 : 3.1Mb/27.5kb). We then used this resource to map the probe sequences of 144 putative SNPs genotyped in 480 individuals. The number of probe-to-genome mappings and alignment length together explained almost a third of the variation in validation success, indicating that sequence uniqueness and proximity to intron-exon boundaries play an important role. The same pattern was found after mapping the probe sequences to the Walrus and Weddell seal genomes, suggesting that the genomes of species divergent by as much as 23 million years can hold information relevant to SNP validation outcomes. Additionally, reanalysis of genotyping data from seven previous studies found the same two variables to be significantly associated with SNP validation success across a variety of taxa. Finally, our study reveals considerable scope for validation rates to be improved, either by simply filtering for SNPs whose flanking sequences align uniquely and completely to a reference genome, or through predictive modelling.