artic_vcf_filter fails

[kclambi1]

Hi! When running Artic on COVID isolates, we randomly seem to get this error only on certain samples:

Running: artic_vcf_filter --medaka XX.merged.vcf XX.pass.vcf XX.fail.vcf
Command failed:artic_vcf_filter --medaka XX.merged.vcf XX.pass.vcf XX.fail.vcf

These same samples are problematic when rerunning, so it's not just a fluke. The samples have good starting material, Ct values, and nothing out of the ordinary for read quality or depth. We aren't sure exactly why this happens or what within these samples causes this error. We are using V3 primers, but have also recently tried V4, Midnight, and recently Varskip together on a replicated sample set. Oddly enough, V3 sets go to completion but V4 and Midnight (still ironing out Varskip) get this error on the same sample. I'm wondering if the fragments from the libraries are causing this somehow? All files but the merged .vcf's and thusly full consensus sequences are generated up to this point.

Just poking to see if anyone else is having this crop up commonly and how they remedied it. There seems to be limited information I can find on this specific failure point. Thanks!