Closed lkhilton closed 2 years ago
I was able to run with an empty file when testing. Can you point it to a dummy file?
Will try and report back, thanks!
If this works can we set it up and document it in the config to make it more clear that users should leave the empty/dummy file unless they replace it with their own?
Absolutely! I did it in demo config, should have thought about the default config as well
# here you can specify path to txt file with a list of custom ENST IDs that override canonical selection
# it will be parsed to --custom-enst flag of vcf2maf
# if no non-canonical transcript IDs to be included, leave switches empty
# This is just an example of how to include the list of custom IDs
switches:
custom_enst:
hg38: ""
grch37: "data/custom_enst.txt"
hs37d5: ""
EDITED: Not even like this, the path should be for each build. I will update it once Laura confirms it worked for her.
This is the result with an empty file:
ERROR: Provided --custom-enst file is missing or empty: etc/dummy_enst.txt
It probably worked for Kostia because he has a copy of that file in his gambl-repo "etc" folder. The config probably just needs to be changed to refer to the one under MODS_DIR or whatever it's called.
it might be the case. However, I also tested it for the demo sample, which is another repo, and another directory. But most likely it worked for me because I have a file there as well. I am setting it up to test outside GSC network, to see if I can come up with the same error.
In my Battenberg config I set the path to src this way:
src_dir: "{MODSDIR}/src/"
I presume this will fix the issue:
grch37: "{MODSDIR}/etc/custom_enst.txt"
Did "data" get substituted with "etc" in a recent commit, perhaps?
I found it a bit unusual but the issue is that the empty file cannot be 0 b. I had a newline (just a return) in my file, so it was 5 b in size and therefore technically not emty. I found that if I remove the newline from a file (making it 0 b), vcf2maf exits with the same error complaining about the empty file. In any case, the switches
in config should not be empty contrary to what I put in the comments, so I will update the module with a patch. As an interim solution, would you mind adding a newline to the dummy file? Thanks for testing it and reporting this issue!
The newline was the key, thanks!
Can this be closed?
This has been resolved. With this patch I wanted to push also update on demo module where vcf2maf is running. I am still working on that update. I will close the issue once issue PR.
I have added documentation for this in the PR https://github.com/LCR-BCCRC/lcr-modules/pull/165
V1.2 of vcf2maf can't be run without a custom_enst value for every genome build. It throws this error:
This is a bit of a circularity problem since you have to first run vcf2maf to find non-canonical transcripts to supply to this argument. Maybe it need an
if
statement to test if the field is empty an run a different command based on that test?