bayesTyperTools combine Assertion `contig_id_to_idx_it != contig_id_to_idx.end()' failed

[martinghunt]

I got this error message:

bayesTyperTools: /isdata/kroghgrp/jasi/bayesTyper/code/releases/v1.3.1_static/BayesTyper/src/vcf++/VcfMetaData.cpp:338: uint VcfMetaData::getContigIndex(const string&): Assertion `contig_id_to_idx_it != contig_id_to_idx.end()' failed.
/ebi/lsf/yoda-spool/02/1529441822.3144105: line 8: 29650 Aborted                 (core dumped) /homes/mhunt/bin/bayesTyper_v1.3.1_linux_x86_64/bin/bayesTyperTools combine -v samtools:../samtools.vcf,cortex:../cortex.vcf -o 03.typertools.combine -z

Maybe I'm not using it as intended?

My use case is a bacterium where I have calls for one sample from samtools and cortex, not from GATK/playpus/Manta. I don't have the "variation prior", just calls for that sample from samtools and cortex, hence no prior:<prior>.vcf in my command line.

Can this type of data be used with BayesTyper?

[lassemaretty]

Hi,

Thank you for posting!

Regarding your specific issue:

bayesTyperTools combine requires the header to contain contig (e.g.##contig=<ID=8,length=146364022>) entries for all reference sequences containing variants in the vcf; the contigs further need to appear in the same order in the header and for the variant entries. We have added information about this to the readme.

Generally about running on bacterial genomes:

We have not tested BayesTyper on bacteria. There are no theoretical reasons why it shouldn't work, but please note that:

• BayesTyper can only call variants where the coverage is an integer multiple (1 or 2) of the haploid coverage. Hence, your sample should probably only contain a single strain.
• BayesTyper currently only uses autosomes (reference sequences not called chrX, X, chrY or Y) to estimates parameters. Hence, while the theoretically optimal way of running it would be to label your reference sequence Y to inform BayesTyper that the organism is haploid, it wouldn't work. For now, you therefore have to run it as diploid (by using a non sex chromosome label) and ignore heterozygote genotypes - hopefully you wont get that many. If you do, it could be a sign that BayesTyper is not working as intended on your data. We will try to push out a release as soon as possible that will estimates parameters also on non-diploid chromosomes and that will support inputting a set of reference sequence ploidies to replace the currently hard-coded sex chromosome plodies.

Please do not hesitate to post again if you further questions or suggestions for how we can improve the BayesTyper support for bacterial data.

br,

Lasse

[martinghunt]

Thank you! I got it working (one of the VCFs didn't have the contig header line)