Hi there,
This page - http://www.bioconductor.org/packages/release/bioc/html/ensemblVEP.html - says that the EnsemblVEP bioconductor package needs v88 of the Ensembl API, but from some of the help pages within ensemblVEP v1.20.0 it looks like maybe v90 is now supported too?
I'm asking because I'm trying to troubleshoot a problem that probably has to do with my installation of the back-end ensembl perl modules and my PERL5LIB variable, rather than bioconductor itself (not sure).
I'm a bit confused, as when I issue what I think is an equivalent unix command-line call to the vep script, it seems to work fine. Does bioconductor make a system call to the vep perl script in order to do the analysis? If so, is there an option within R to show me the command it's trying to run, so I can replicate that and and test it from the command line? Where does bioc take PERL5LIB from (on a linux box) - is it simply taken from my environment when R starts up?
thanks very much,
Janet
Thanks, Lori - that did let me troubleshoot successfully. I did just put in a github issue with a request to see the system call - that was really useful once you showed me how to do it using debugonce.
I should probably update to VEP v90 soon, but perhaps it's also useful to let you know where I stumbled: it seems simple to build in a test within VEPParam that would prevent this mysterious error.
It turns out that the default host used by the bioc module useastdb.ensembl.org) does not support anything other than the current or current minus 1 versions (according to https://www.ensembl.org/info/data/mysql.html), and gives a very uninformative error when you try to use an older version of the vep script, whether on command line or through R/Bioc. The default host for the command-line vep script is different ensembldb.ensembl.org) - that's why I didn't replicate the error on the command line. Using ensembldb.ensembl.org as host instead fixes the issue.
I've given details below - does that work for you, or should I put this into a github issue?
So, while this fails with a mysterious error (that was what made me suspect my vep/perl installation):
when I specify a different host it works fine:
Would it be possible to change the default host when older versions of VEP are specified? I didn't expect that to be an issue, because on my real data I was using a local cached database and didn't think I was contacting any host at all. The same fix works on my data, though, so I'm happy:
Session info is below (truncated - does github really only allow 5000 chars in a reply?). Thanks for the quick response - it's nice to have this figured out already.
Janet
---------
Thank you! also very informative. If you could also open an issue for this. We do use the github repositories now for tracking issues and that way it stays there until it is fixed and doesn't get lost in the support site issue. Just please copy the code chunks you provided and if you could include a link to this support site issue that would be wonderful.
Glad you were able to get it straightened out.
Cheers,
Lori
done (https://github.com/Bioconductor/ensemblVEP/issues/2). thanks again!