Regarding deleted SNPs (in dbSNP) due to mapping or clustering errors found in VCF file

Question

[CrazyB]

Need some input from the community concerning the question in title -

Some of the SNPs deposited in dbSNP database were later removed due to mapping or clustering errors - for example rs11273140 (which you would find nothing using NCBI or Ensembl, one record in 1000 genomes browser, which states the rs is deleted).

If this rs is removed in dbSNP due to mapping/clustering errors, what does it mean when my VCF file reports such a variant in my sequencing cohort ?

(a) there is a mapping/clustering error during my sequence alignment (not using an updated ref seq? not using an updated dbSNP list? other possibility? ) (b) sequencing platform issue (e.g. exome seq v whole genome seq seem to report different SNPs) (c) others?

Any insight/reference will be appreciated. Thank you.

Answer 1

[Ying W]

So what usually happens is a VCF is generated by a variant caller and then these VCFs are then annotated. Annotation software will usually have some database (like dbSNP) that was downloaded locally that is used for annotation. These databases are not updated very often.

What you are seeing is that one of your VCFs is annotated with an rsID that is deleted in the most recent version of the annotation database but the most recent version is not what was used to annotate your VCF. This has nothing to do with your mapping/aligning and is not a platform issue. I don't think you should worry and can just ignore the that rsID.

[CrazyB]

Thanks a lot, Ying. A follow-up (?) question: Regardless of whether local dbSNP is updated or not, if a VCF reports a finding, doesn't it mean by definition at that particular genomic position, there exist two or more variants. So if updated dbSNP removes the original rsID, would that make this "new discovery" a novel SNP ???