[ensembl-dev] VEP + ExAC plugin run on GRCh38: website finds ExAC data, command line does not?

[Vivek Iyer]

I’m trying to run VEP with the exac plugin over GRCh38:

I’ve isolated a little test-set for a gene (SERPIND1) where I can see Exac variants. 
The actual variant in GRCh38 is:
22	20779735 20779735 G/A

(this is mapped to GRCh37:22:21134023-21134023, which has a lovely synonymous ExAC variant)

When I put this variant into the GRCh38 Ensembl web interface, it works a treat and recovers the exac data for the single consequence (AF=0.02). Hooray, go ensembl.

However when I run via the command line (I include a small.test2.vcf with this single variant):

perl /software/vertres/bin-external/variant_effect_predictor_v82.pl -offline --vcf --species homo_sapiens --assembly GRCh38 --no_progress --everything --cache --pick --dir_cache /data/blastdb/Ensembl/vep  --chr chr22 -o small.test2.vep.vcf --force_overwrite -i small.test2.vcf --verbose --plugin ExAC,/lustre/scratch116/casm/team113/ref/exac/exac.r0.3/ExAC.r0.3.sites.vep.vcf.gz

- I get no ExAC annotation applied to this single consequence (the last 8 entries of the CSQ field).

I notice the exac vcf file I put in (indicated via the VEP docs) is the published GRCh37 file from broad.

Did ensembl produce a 38 remap? The implication is that this should work fine for GRCh38 but do I have to lift over the ExAC file myself?

Cheers,

Vivek

-------------
Vivek Iyer
Experimental Cancer Genetics
Wellcome Trust Sanger Institute
vvi at sanger.ac.uk
+44-1223-495364






-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ensembl.org/pipermail/dev_ensembl.org/attachments/20151104/2737bcc1/attachment.html>