Understanding variants that PASS filter but are also flagged?

Question

[rtho]

I'm working with some VCF files from a collaborator. In the FILTER column of any given VCF I'm used to seeing either PASS or some form of flag; however, in the files I'm looking at now, a reasonable proportion of variants are annotated PASS BUT also have one or more flags e.g. 'alt_allele_in_normal,clustered_events,t_lod_fstar,PASS,homologous_mapping_event,triallelic_site'

Is that normal, or does this suggest something went awry with the raw data processing (I'm awaiting details on the methodology used)? I've never seen this before. Thanks

[rtho]

Thanks- I can see how a given variant could fail for multiple reasons, but the part that doesn't make sense to me is the fact that these variants are ALSO annotated with 'PASS'

i.e. in the above, example 'alt_allele_in_normal,clustered_events,t_lod_fstar,PASS,homologous_mapping_event,triallelic_site'

If it fails QC, why is it also marked PASS?

Thanks

[Kevin Blighe]

If it fails QC, why is it also marked PASS?

Please ask your collaborator how they processed the data. Like I said, the FILTER field in the VCF may not even be used by them (the collaborators), in which case the software calling the variants will decide what passes or fails based on its own default settings.

Answer 1

[Kevin Blighe]

I would say that this is common - yes. Variants can fail QC in different ways.

You should really ask your collaborator how the value in the FILTER column has been set, though. It could be that they are not even utilising it, in which case it will be set by whatever program has called the variants and using it's default settings (but even the program may not even do this). However, if the collaborator is utilising it, then it may represent just one very fundamental aspect of QC.

Either way, you should check.

Kevin