Are 100,000 "SNPs" Useless?

Bailey et al. (1) used public and private genome sequences to define segmental duplications within the human genome. Their excellent study demonstrated that the 5.2% of the genome present as segmental duplications contains 6.1% of known exons and roughly 100,000 more single nucleotide polymorphisms (SNPs) from the public SNP database (dbSNP) than expected. This latter feature led the authors to conclude that “about 100,000 paralogous sequence variants currently contaminate dbSNP.” In other words, these entries in dbSNP do not represent allelic variants (polymorphisms), but differences between paralogous sequences (cismorphisms). This calculation assumed that “there is no reason to expect that polymorphic variation is increased within duplicated regions.”