Given a rank model with
[most_severe_consequence]
category = consequence
data_type = string
description = Most severe consequence for this variant
field = INFO
csq_key = Consequence
info_key = CSQ
record_rule = max
separators = ',', ':', '|',
genmod can properly rank non_coding_transcript_variant higher than intron_variant from a intron_variant&non_coding_transcript_variant consequence, given that the priority of non_coding_transcript_variant is higher than intron_variant.
This is great! But it should be documented (either here or e.g. in reference-files) that this works out of the box (that an ampersand is not needed a separator) and should (perhaps) be the recommended way of writing a rank model.
Other scripts used in pipelines have for example disregarded everything after the ampersand, and only considered the first consequence.
Given a rank model with
genmod can properly rank
non_coding_transcript_varianthigher thanintron_variantfrom aintron_variant&non_coding_transcript_variantconsequence, given that the priority ofnon_coding_transcript_variantis higher thanintron_variant.This is great! But it should be documented (either here or e.g. in reference-files) that this works out of the box (that an ampersand is not needed a separator) and should (perhaps) be the recommended way of writing a rank model.
Other scripts used in pipelines have for example disregarded everything after the ampersand, and only considered the first consequence.