Most SV types missing from genotyped vcf

[afurches]

Hi,

I'm using MetaSV to merge and genotype BreakDancer calls (CTX,DEL,INS,INV,ITX) across multiple contigs and subpopulations. To replicate what I'm doing with other software, I want to use MetaSV to:

1. merge SV calls across groups and contigs,
2. use the merged vcf file to genotype all samples individually,
3. merge all genotyped vcf files into a multisample vcf.

Step (1) seems to work well and quickly - I'm able to get breakdancer.vcf.gz and variant.vcf.gz output files, and all SV types are present.

$ zcat breakdancer.vcf.gz | grep -v '#' | cut -f5 | sort | uniq
<CTX>
<DEL>
<INS>
<INV>
<ITX>

$ zcat variants.vcf.gz | grep -v '#' | cut -f5 | sort | uniq
<CTX>
<DEL>
<INS>
<INV>
<ITX>

But when I genotype samples using either merged vcf (--breakdancer_vcf), only DEL calls are genotyped/reported, even if I specify --svs_to_report {INV,CTX,INS,DEL,ITX}.
For other SV types, the error files say Skipping Record(...) due to small size.

EXAMPLE:

...
ERROR 2024-12-06 13:02:29,571 metasv.sv_interval   Ignoring record due to missing SVTYPE or INFO field in Record(CHROM=genome_chloro|Contig01+38342752_1, POS=82170, REF=G, ALT=[<INS>])
WARNING 2024-12-06 13:02:29,571 metasv.sv_interval   Skipping Record(CHROM=genome_chloro|Contig01+38342752_1, POS=83010, REF=T, ALT=[<ITX>]) due to small size
WARNING 2024-12-06 13:02:29,571 metasv.sv_interval   Skipping Record(CHROM=genome_chloro|Contig01+38342752_1, POS=84336, REF=T, ALT=[<ITX>]) due to small size
ERROR 2024-12-06 13:02:29,571 metasv.sv_interval   Ignoring record due to missing SVTYPE or INFO field in Record(CHROM=genome_chloro|Contig01+38342752_1, POS=84903, REF=A, ALT=[<INS>])
ERROR 2024-12-06 13:02:29,572 metasv.sv_interval   Ignoring record due to missing SVTYPE or INFO field in Record(CHROM=genome_chloro|Contig01+38342752_1, POS=104631, REF=A, ALT=[<INS>])
INFO 2024-12-06 13:02:29,572 metasv.main          SV types are set(['DEL'])
INFO 2024-12-06 13:02:29,572 metasv.main          Do merging
INFO 2024-12-06 13:02:29,572 metasv.main          Processing SVs of type DEL
INFO 2024-12-06 13:02:29,572 metasv.main          Intra-tool Merging SVs of type DEL
INFO 2024-12-06 13:02:29,572 metasv.main          First level merging for DEL for tool BreakDancer
INFO 2024-12-06 13:02:29,902 metasv.main          Inter-tool Merging SVs of type DEL
INFO 2024-12-06 13:02:30,061 metasv.main          Output merged VCF without assembly
INFO 2024-12-06 13:02:30,313 metasv.main          ('DEL', 'LowQual', 'IMPRECISE', ('BreakDancer',)):572
INFO 2024-12-06 13:02:30,342 metasv.main          Clean up pybedtools
INFO 2024-12-06 13:02:30,342 metasv.main          All Done!

# genotyped output
$ zcat variants.vcf.gz | grep -v '#' | cut -f5 | sort | uniq
<DEL>

Why are all SV types are present in the merged vcfs, but only DEL are genotyped?

I'll plan on using a different software to genotype and merge the calls for now, but wanted to ask whether this is a bug.

Thanks,
A