POPULATIONS Unable to Read Whitelists Produced in VCFtools

[Jack Dirck]

Hi all!

I am using STACKS (v2.65 ) and VCFtools (v0.1.16) to filter my SNP dataset of 480 individuals (329 after filtering for missing data). I assembled the loci de novo in the STACKS pipeline. I filtered missing data iteratively in PLINK and used lists of removed loci and individuals as blacklists in populations. 

I switched to filtering in VCFtools for minor allele count and depth filtering. I would stick with VCFtools but I would like to use the --write-single-snp command in populations for LD pruning. However, I am running into problems when moving from VCFtools to populations. My strategy has been to filter in VCFtools, use the --kept-sites command to make make whitelists for populations to use. However, every time I do this populations fails to remove any loci not on the whitelist even though it reads it properly. I understand that VCFtools and populations number loci differently and I have modified my whitelists accordingly but that does not work either. I know the formatting is correct for my whitelists and populations does not have an issue parsing the files. 

Any ideas on what might be going on? Thank you!

Cheers,
Jack Dirck

Code and log files attached below: 

[Catchen, Julian]

Hi Jack,

 

You didn’t include a subset of your whitelist, so I could not evaluate its format. A couple points: once you export the data as a VCF file of SNPs, it is no longer associated with its RAD locus, so you cannot use --write-single-snp when you import, since Stacks sees all the VCF-imported SNPs as independent loci. You can, however, export the data with --write-single-snp prior to doing your filtering. Second, in Stacks v2.67 we exposed a minimum depth filter, which you can use directly in populations with --min-gt-depth.

 

Best,

 

Julian

[Jack Dirck]

Hi Julian,

Thank you so much for your feedback. I think that cleared it up

Thank you!

Cheers,

Jack