bcftools 1.19 - consensus - different warnings for overlapping variants

[salvatoreloguercio]

Hello,

I am running bcftools consensus on a series of genomic regions (fasta), by applying variants from a single subject and getting IUPAC codes as output ( -s -H I ).

For overlapping variants, now I am getting two kinds of warnings - 1) "The site chr[..]:[..] overlaps with another variant, skipping..." and 2) "Warning: ignoring overlapping variant starting at chr[..]:[...]"

More specifically, suppose I have two regions - chr1:100-200, and chr1:300-400 (just using these small coords as example):

• bcftools consensus run on the first region will output only warnings of type 1).
• when running consensus on the second region, I get the warning 2) for positions belonging to the first region (chr1:100-200), then the usual warning 1) for positions within the actual region being processed (chr1:300-400).

In this example, the vcf used for both regions would be the same and have variants spanning both intervals (chr1:100-400).

I am used to the warning 1) since previous versions of bcftools, but kind of confused by the warning 2) - since it considers earlier variants.

Just wondering if you have an explanation for this behavior - hopefully not a bug?

[pd3]

The two warnings are emitted from two slightly different contexts. The code is a bit opaque, so it's hard to tell without seeing specific examples, but the difference should be harmless.