sniffles 2.5 produces variant longer than contig

[bbimber]

Hello,

I ran sniffles 2.5 to produce a merged VCF, after processing 59 PacBio CLR and CCS datasets with sniffles. I ran this with default params:

$SNIFFLES \
	--no-progress \
	--input fileList.tsv \
	--allow-overwrite \
	--vcf $VCF

The resulting VCF has a variant on contig QNVO02001917.1 (an unplaced contig from the MMul_10 genome), which starts at 51115 and would end at 57949; however, the contig is 57948 bp long. GATK and similar tools complain about it being non-compliant. I attached the variants from this contig in a text file. Should this occur?

QNVO02001917.1.txt

[bbimber]

I have a second similar site from the same dataset:

QNVO02002480.1	594	Sniffles2.DUP.2M174A	N	<DUP>	34	PASS	PRECISE;SVTYPE=DUP;SVLEN=5142;END=5736;SUPPORT=1;COVERAGE=0,4,13,10,3;STRAND=+-;AC=6;STDEV_LEN=0.806;STDEV_POS=292.300;SUPP_VEC=10101010001001000000101100000010000001000010100000001000000

All variants from that contig are here:

QNVO02002480.1.txt

[bbimber]

Hello - just wanted to check back and see if anyone has comments on this. While we can manually drop these invalid SVs, it would be very helpful if sniffles did not output them to begin with.

[fritzsedlazeck]

So you pointing at that the end position is +1 bp too long? @hermannromanek we can put a cross check in the output.
Thanks
Fritz

[bbimber]

Yes, 1bp longer than the reference length

[lfpaulin]

@hermannromanek check if POS and END are 0-based or 1-based
we had had discussion some months ago with a file having a POS=0