We used three different tools for functional annotation analysis. To distinguish between variants with functional effects (i.e., variants that affect gene regulation and/or protein activity) and variants with no regulatory or activity effects, we used VARIANT (VARIant ANalysis Tool) [30]. Based on the information obtained from VARIANT, we distinguished: variants likely to be nonfunctional (i.e., variants mapping to putative nonfunctional regions), variants with low potential for functional effect (i.e., variants with only a generic annotation: ‘located in a regulatory region annotated by Ensembl’) and variants with high potential for functional effect (i.e., variants with a specific annotation: ‘CpG island’, ‘miRNA target site’, ‘splice site’, ‘splice donor variant’, ‘RNA polymerase promoter’, ‘transcription factor binding site’, ‘splice acceptor variant’, ‘non-synonymous variant’ or ‘stop codon’). In the annotation analysis performed by VARIANT based on in silico evidence, we considered only information related to transcripts annotated in the Consensus Coding Sequence (CCDS) database. We also used rSNPBase (database for curated regulatory SNPs) and RegulomeDB [31,32] to further investigate variants potentially associated with epistatic effects on SD risk alleles. Both rSNPBase and RegulomeDB perform functional annotation on the basis of in silico and experimental evidences.
