We have focused on removing nonconsensus introns, those not matching the known splicing patterns of GT..AG, GC..AG, and AT..AC. We found ∼2200 such introns in GENCODE 7, which is a reduction from 3c, where there were ∼3300 nonconsensus introns. For version 7, 13% of these show good cDNA support. A small number of nonconsensus splice sites are believed to be other donor/acceptor combinations recognized by U12 spliceosomes. Others are suspected polymorphisms, some with SNP support. Only 2% could be converted to canonical splice sites by known SNPs from dbSNP version 132 in some members of the population, but this proportion has doubled since 3c, reflecting the recent increase in SNP discovery from genome-wide projects. There may be more of these splice sites that are polymorphic but the SNPs are as yet unknown. There are a number of nonconsensus splice sites that are only one base different from canonical, suggesting that they could have been formed due to mutation, but some may represent low frequency polymorphisms. As the 1000 Genomes Project (http://www.1000genomes.org) and other human sequencing projects progress, more SNPs will be discovered, and therefore, the number of known polymorphic splice sites should increase.
