Hello everyone, I have been running nf-core/sarek 3.9.0 using the hg19 reference genome and the SnpEff GRCh37.87 database for variant annotation. I noticed that mitochondrial variants were not annotated correctly. In the hg19 reference genome, the mitochondrial chromosome is named

chrM

, whereas in the SnpEff GRCh37.87 database it is named

MT

. This chromosome-name mismatch appears to prevent SnpEff from annotating the mitochondrial variants. Using the SnpEff

hg19

database resolves the chromosome-name mismatch, but the resulting annotations appear to be less comprehensive than those obtained with GRCh37.87 for the other chromosomes. Has anyone encountered this issue? What would be the recommended way to annotate mitochondrial variants while retaining the more comprehensive GRCh37.87 annotations?

I haven't used snpEff much, it doesn't seem very well maintained (seems stuck in version 99, which I think corresponds to the Ensembl release that came out Jan. 2020, just before the pandemic). I prefer VEP, and VEP actually has a file in its cache that translates contig names (e.g.

MT

,

chrM

,

NC_012920.1

,

J01415.2

are "synonyms"), so maybe that would work better for your use case? That said, the

GRCh38

versions of snpEff (looked at

86

and

99

) seem to use

chrM

(and they work fine with sarek). I download the cache using the snpEff container used in sarek with the command

snpEff download "$genomeversion" -dataDir "$dir"

. Is snpEff

GRCh37.87

different, or does it matter how you download the cache?

In case someone is interested in the future, I have found this solution