Insertion to Duplication post-process bug?

[vkaz39]

Hello there,

I'm writing to report a couple of issues I've encountered while using Jasmine. I have used six tools independently to call SV on the same sample, say sample1, as well as pre-processed calls to convert duplications to insertions and corrected errors in insertions with Iris.

After per-sample merging with Jasmine, I observed the following unexpected result when running the bcftools norm command (using --check-ref e):

Reference allele mismatch at CM018180.1:14965491:
REF_SEQ: 'T'
VCF REF: 'N'
After performing post-processing (converting insertions back to duplications) with Jasmine, I encountered another issue based on the same bcftools norm command:

Non-ACGTN reference allele at CM018180.1:3761701:
REF_SEQ: 'C'
VCF REF: '.'

I've also verified the original VCF files and the output VCF from Iris tools, and they do not show any discrepancies between the reference sequence and the VCF REF.

When I checked the code within Jasmine (InsertionsToDuplications.java), I see that setRef is '.' for setAlt "<DUP>". Why is it so?

if(line.contains("OLDTYPE=DUP") && ve.getType().equals("INS"))
			{
				countDup++;
					
				long start = ve.getPos();
				int length = ve.getLength();
				long nstart = start - length + 1, nend = nstart + length;
				String refinedAlt = ve.getAlt();
				ve.setPos(nstart);
				ve.setInfo("END", nend+"");
				ve.setType("DUP");
				ve.setInfo("REFINEDALT", refinedAlt);
				ve.setInfo("STRANDS", "-+");
				ve.setRef(".");
				ve.setAlt("<DUP>");
				out.println(ve);
			}
			else
			{
				ve.setInfo("REFINEDALT", ".");
				out.println(ve);
			}

Could you please let me know if these issues indicate a bug in the Jasmine?

Thank you!
Niraj