Repository navigation
Why is the conservation score for conserved splice variants 0? #39
Description
Hi!
I noticed that a homozygous canonical splice variant got a relatively low rank score, and wanted to understand why.
The variant has rank score 23, so high enough that we find it, but still surprisingly low. It's this variant:
https://scout38.sys.scilifelab.se/cust003/26328/ee08cf6c379a052bf693e8e8776bc51b
The splice variant has conservation = 0 in rank score. But it appears to be quite conserved when I look in UCSC. PhyloP [100 vertebrates] is 6.5 (max 10), PhastCons is 1 (max value). I did a liftOver to hg19 to see the GERP score, which is 6.07 (near max).
I checked a couple of other splice variants in hg38, to see if they also get a conservation score of 0. I looked at around 10, and all had conservation score = 0. Some are not conserved, so for those it makes sense. But these variants all have a conservation score = 0 but look quite conserved in UCSC:
https://scout38.sys.scilifelab.se/cust003/26306/a556dbfe8674dc3f8902a2f4903726b2
https://scout38.sys.scilifelab.se/cust003/26302/d576dffe9a7c2d219b5323f0b42a0d04
https://scout38.sys.scilifelab.se/cust003/26302/def0324ee7fd029ea2e24206a38b1edd
https://scout38.sys.scilifelab.se/cust003/26328/e0b931f80d8282c178e7171328117e77
I only came across one splice variant that has a conservation score > 0. This variant looks less conserved (PhastCons track in UCSC) than the examples given above, but it is actually a missense variant. I.e. consequence = splice variant and consequence (MANE transcript) = missense variant.
https://scout38.sys.scilifelab.se/cust003/26305/a04c0eb6b79fc37c16966ab1d9b3c9b0
I'm wondering if all splice variants, including conserved canonical splice variants, get a conservation score of 0 (unless they are missense variants in some transcripts). I don't think that's your intention? I don't know if splice variants were treated the same in hg19.
I'm looking forward to hearing your thoughts!
Activity
jemten commentedon Aug 26, 2026
Thank you for bringing this up! These values comes from the pipeline and we will have to check as you say if something has been missed in the shift to the new reference genome.
northwestwitch commentedon Aug 26, 2026
Thanks @jemten, I'll also take a look if anything went wrong parsing the VCF file
dnil commentedon Aug 26, 2026
Thank you all! I had a quick look, and did not see any conservation values on the first variant in the VCF, so will also assume this is a pipeline annotation/reference file issue. It is on some level also comforting to see some variants with conservation at least, making this less likely to be a Scout parsing issue. Unless you find something @northwestwitch, maybe we move this to some open pipeline repo?
northwestwitch commentedon Aug 26, 2026
Nope, same conclusion here.
sofiademmou commentedon Sep 7, 2026
Two issues identified for these splice variants :
pep_match=0) prevents the plugin from dropping these lines. It should now retrieve positional annotations (GER P++, phyloP, phastCons) using the genomic position.By default, the plugin only runs for:
missense_variant,stop_lost,stop_gained, andstart_lost. Addingconsequence=ALLshould also retrieve these splice variants.