Change parameter for VEP annotation of CNVs
Bakground
Current VEP annotation of CNVs uses a parameter --per_gene
that selects a single transcript per gene, which is annotated with consequence and shown in ELLA's side and top bars. However, --per_gene
uses a set of criteria that includes status as "canonical" transcript as one of the most important. This produces confusing results for interpreters, as Ensembl's definition of "canonical" not always matches the gene panel transcript (MANE Select, which is not counted by VEP for hg19).
In addition, unlike for SNV interpretation, no meaningful "worse consequence" warning is possible, as only one transcript is annotated. This can potentially result in interpreters overlooking a deleterious effect in an important transcript (default or other).
From comments, the setting appears to simply have been chosen from SciLifeLab's default settings for FindSV.
Implementation
Alternatives:
- Remove
--per_gene
parameter to annotate variants on all RefSeq transcripts (as for SNVs). Note that this might produce unwanted side effects, especially for larger CNVs, so need to be tested properly. - Change to showing transcript with "worst consequence" only by changing the parameter
--per_gene
to--per_gene --pick_order rank
in/targets/modules/wgs_cnv_annotation.sh
.
Suggest to test both options on one or more suitable test samples and inspecting the results in ELLA before deciding.