--strict fails about artic-tools HOT 3 CLOSED

Thanks for posting the issue here - I'll take a look. Hopefully before the end of the week!

from artic-tools.

OKay so did some digging, I'm using the default nanopolish option, and the file consumed at this stage SAMP_ID.merged.filtered.vcf looks to be missing the canonical vcf spec ##contig= line specifically the VCF header looks like this:

##fileformat=VCFv4.2
##nanopolish_window=MN908947.3:1-29902
##INFO=<ID=TotalReads,Number=1,Type=Integer,Description="The number of event-space reads used to call the variant">
##INFO=<ID=SupportFraction,Number=1,Type=Float,Description="The fraction of event-space reads that support the variant">
##INFO=<ID=SupportFractionByStrand,Number=2,Type=Float,Description="Fraction of event-space reads that support the variant for each strand">
##INFO=<ID=BaseCalledReadsWithVariant,Number=1,Type=Integer,Description="The number of base-space reads that support the variant">
##INFO=<ID=BaseCalledFraction,Number=1,Type=Float,Description="The fraction of base-space reads that support the variant">
##INFO=<ID=AlleleCount,Number=1,Type=Integer,Description="The inferred number of copies of the allele">
##INFO=<ID=StrandSupport,Number=4,Type=Integer,Description="Number of reads supporting the REF and ALT allele, by strand">
##INFO=<ID=StrandFisherTest,Number=1,Type=Integer,Description="Strand bias fisher test">
##INFO=<ID=SOR,Number=1,Type=Float,Description="StrandOddsRatio test from GATK">
##INFO=<ID=RefContext,Number=1,Type=String,Description="The reference sequence context surrounding the variant call">
##FORMAT=<ID=GT,Number=1,Type=String,Description="Genotype">

I think valid looking contig line in the header should look like this:

##contig=<ID=MN908947.3,length=29903,URL=https://raw.githubusercontent.com/artic-network/primer-schemes/master/nCoV-2019/V3/nCoV-2019.reference.fasta,species="Severe acute respiratory syndrome coronavirus 2">

I guess if nanopolish is running, details from the primer scheme ref file can be used to complete these details and make sure at very least the ID line and length are completed. Which should hopefully get vcf_parse working.

from artic-tools.

Looks like the problem was an unindexed VCF. It didn't seem to be an issue in my older testing environment but appears with the latest env release - my bad for not testing this feature properly before deploying!

Closing this issue in favour of the fix to the artic pipeline over in artic-network/fieldbioinformatics#68 .

This leaves artic-tools requiring an indexed VCF and adds the required indexing step to the artic pipeline. Decision for this is because the indexed VCF will be useful later on when I get round to adding more features to the vcf checker.

Thanks again for raising this issue @MattBashton!

from artic-tools.