galaxyproject / tools-iuc

Tool Shed repositories maintained by the Intergalactic Utilities Commission
https://galaxyproject.org/iuc
MIT License
160 stars 425 forks source link

report a bug: Jbrowse can not find biopython #3667

Closed songshuang1996 closed 3 years ago

songshuang1996 commented 3 years ago

Dear developers: I want to use Jbrowse on galaxy, however it is not working.The error is as below: Traceback (most recent call last): File "/Users/songshuang/galaxy/database/shed_tools/toolshed.g2.bx.psu.edu/repos/iuc/jbrowse/b5c5470d7c09/jbrowse/jbrowse.py", line 16, in from Bio.Data import CodonTable ModuleNotFoundError: No module named 'Bio'

I found many other had the same problem. (https://biostar.usegalaxy.org/p/21815/index.html).And I tried many version of Jbrowse on galaxy(1.16.11+galaxy1,1.16.5+galaxy7,1.16.1+galaxy0,0.7.0), even reinstalled galaxy, however either of them failed.

How can I fix this problem? Thank you.

bernt-matthias commented 3 years ago

@songshuang1996 can you check in the Galaxy admin interface under Manage Dependencies if the dependency for jbrowse are resolved with conda (there should be a green checkmark).

songshuang1996 commented 3 years ago

Thank you for reply. Yes, I checked dependencies. Everything looked fine(python 2.7 Jbrowse biopython etc), with green checkmark on the right. Sorry I left my computer in my office, thus I can not provide screenshot of that page now. I will provide it as soon as possible.

songshuang1996 commented 3 years ago

Screen Shot 2021-05-14 at 8 35 07 AM

bernt-matthias commented 3 years ago

Can you post the content of tool_script.sh from the job's working dir?

bernt-matthias commented 3 years ago

If the working dir is purged you also find it in the logs?

songshuang1996 commented 3 years ago

The tool_script.sh only appear in a flash, and then removed automatically with the folder(like rm -r ~/galaxy/database/jobs_directory/000/20) containing it. This might be the working dir you mentioned: ~/galaxy/database/jobs_directory/000/20/tool_script.sh tool_script.txt

bernt-matthias commented 3 years ago

Set cleanup_job to onsuccess in galaxy.yml and restart. Then the working dir of failed jobs won't be deleted. Then we need the content of the tool_script.sh.

songshuang1996 commented 3 years ago

Yes, I have post it in last comment. That file I renamed to tool_script.txt.

bernt-matthias commented 3 years ago

Can you

bernt-matthias commented 3 years ago

Do you have any conda or python virtual env activated by default?

songshuang1996 commented 3 years ago

I do have a base conda env activated. base_conda_list.txt After I activate '/Users/songshuang/galaxy/database/dependencies/_conda/envs/mulled-v1-8caab3d6756f1460da0538aac50a0ca397d7e8b4a8ff79d7d838af7830f44e58' galaxy_conda_list.txt And execute the last commands in the tool_script.sh. The commands seemingly work well. The output is: 'WARNING: No feature names found for indexing, only reference sequence names will be indexed.' Screen Shot 2021-05-16 at 8 14 57 PM

And my computer is M1 mac. But I think it will not influence conda environment. Hope these information can be helpful. Thank you.

bernt-matthias commented 3 years ago

Can you try to configure conda to not activate base by default and try again to execute via galaxy?

songshuang1996 commented 3 years ago

It works when deactivating conda environment. Thank you. And sorry for disturbing you. I have one more question, I face the same error when using Jbrowse on my local galaxy and https://usegalaxy.org/. Screen Shot 2021-05-17 at 4 03 40 PM Is there any way to fix this problem? Thank you.

bernt-matthias commented 3 years ago

Wonderful. I think for the other problem the following helps:

The reason is that Galaxy forbids to show html output of tools by default.

I'm going to close this issue. Please feel free to reopen if needed.

mjbludwig commented 2 years ago

I am having the exact same symptoms but I do not think I have a base environment being loaded? The environment is definitely built correctly since importing works fine on the command line with it loaded.

I can only imagine its due to the same issue of having a base conda env active but I am not sure where its getting that from?