A detailed characteristics of bias associated with long runs of homozygosity identification based on medium density SNP microarrays

摘要

In the present study, runs of homozygosity (ROH) detected with the use of a standard bovine 54ksingle nucleotide polymorphism (SNP) genotyping assay and two different ROH detectionapproaches, based on 50 (M1) or 15 (M2) consecutive SNPs, were compared with results of wholegenome sequencing. Both microarray-based methods accurately recognised medium-sized ROH,however, it was found that M2 method seemed to better than M1 identify short ROH, but highlyoverestimated their number, leading to numerous false positive calls. Moreover, long ROHidentified with microarray data tended to break into shorter segments in sequencing data because ofthe presence of regions with high heterozygosity within the ROH sequences. This may indicate, thatthese long ROH are formed by closely positioned shorter homozygous segments that may be ofolder origin or may be created by two similar but not identical haplotypes, showing minor internalrecombination signs. Such finding also suggests that at least some of the results of previous studiesin regard to long ROH may be biased leading to inaccurate estimations of genomes autozygosity viaROH classification into length categories.