voila tsv error

[Alexander Dietrich]

Hi everyone,

I want to use the voila tsv command, but I am running into a strange error, maybe someone can help me?

2021-01-09 19:34:13,874 (PID:837) - INFO - Command: /usr/local/bin/voila tsv /MOUNT/output/majiq-output/build/splicegraph.sql /MOUNT/output/majiq-output/psi/ALLBAMS.psi.voila --file-name /MOUNT/output/test.out
2021-01-09 19:34:13,874 (PID:837) - INFO - Voila v2.0.0-05c9c32
2021-01-09 19:34:13,874 (PID:837) - INFO - config file: /var/tmp/tmphqugvdfl
2021-01-09 19:34:13,880 (PID:837) - ERROR - option values must be strings
Traceback (most recent call last):
  File "/usr/local/lib/python3.6/dist-packages/voila/run_voila.py", line 166, in main
    config.write(args)
  File "/usr/local/lib/python3.6/dist-packages/voila/config.py", line 185, in write
    config_parser.set(settings, 'analysis_type', analysis_type)
  File "/usr/lib/python3.6/configparser.py", line 1192, in set
    self._validate_value_types(option=option, value=value)
  File "/usr/lib/python3.6/configparser.py", line 1177, in _validate_value_types
    raise TypeError("option values must be strings")
TypeError: option values must be strings

Cheers and thanks for any help,

Alex

[Paul Jewell]

Hi Alex, 

I'm unsure of this specific error, but I do have one suggestion. I don't see the long format "--file-name" argument available in the documentation (https://biociphers.bitbucket.io/majiq/VOILA_tsv.html) , and additionally, the list of files is recommended to be put as the last arguments, so that other arguments may not be parsed as files. Could you try like this?

/usr/local/bin/voila tsv -f /MOUNT/output/test.out /MOUNT/output/majiq-output/build/splicegraph.sql /MOUNT/output/majiq-output/psi/ALLBAMS.psi.voila

I would also generally recommend installing voila in a virtualenv instead of system-wide if possible. 

Thanks, 

-Paul

[Alexander Dietrich]

Hi Paul,

i installed majiq in a docker container, thats why i used this version. I also can run majiq just fine, i only struggel with voila.

I tried your recommendation, but i get the same error.

Best,

Alex

[Alexander Dietrich]

I also tried to wrap my arguments in "", since the error message is saying something that the options have to be strings, but that also did not work.

[Paul Jewell]

Hi Alex, 

I'm looking over the logic surrounding that section where the error is occurring. I'm seeing that the only way the analysis type might not be a string type is if the voila file was not saved with a string type, or became corrupted. How large are the psi file and splicegraph? Would you object to sending them to be to take a quick look? Another possibility might be to try just running $ majiq psi one more time, with a different output file, to see if that might make a difference and rule out corruption. 

Thanks, 

-Paul

[Alexander Dietrich]

Hi Paul,

here are the two files i used: https://drive.google.com/drive/folders/1jS6R-gjrMJBFPNSjvTYd3iWuIjtDU4FC?usp=sharing

I also ran majiq again to create those files, but it still did not work.

I then installed majiq outside of the docker on my local machine and  it worked with those files, so i suspect something is going wrong with the installation. I added the Dockerfile I used as well, maybe you see the issue there?

The thing that confuses me the most is that i got voila tsv to run ~ 1 month ago with this exact Docker.

Best,

Alex

[Paul Jewell]

Hi Alex, 

I've ran with your files on my local standard venv-based voila installation, and it worked without error. (I've attached the output). To that would seem to rule out issues with the files / generation. Can you explain to me what setps you have used to set up the voila docker environment? I'm not as familiar with this method but I may be able to spot a mistake and help to resolve it, or at least find a version that I can get to work correctly on my machine using docker. 

Thanks, 

-Paul

[dietrich.al...@gmail.com]

Hi Paul,

I can walk you through the Docker steps here:

1) install htslib 1.9

2) install python 3.6 and upgrade pip to 3.6

3) pip3.6 install wheel, setuptools, numpy, cython

4) clone majiq repo: https://bitbucket.org/biociphers/majiq_stable.git

5) pip3.6 install requirements.txt

6) python3.6 setup.py install

Thats basically it. Sine majiq works without any problems, I don't know why voila has this issue.

Thanks,

Alex

[dietrich.al...@gmail.com]

Hi Paul,

i have discovered, that the value of "analysis_type" in this line: https://bitbucket.org/biociphers/majiq_stable/src/05c9c32ce02c63d6876b45433affa5ef47382fdd/voila/config.py#lines-185

is in my case of class <class 'bytes'> with value b'psi' and not a string. Maybe this leads to the error message that I get? Do you have an idea how to fix that?

Cheers,

Alex

[Paul Jewell]

Hi Alex, 

The majiq_stable version of the repository is no longer maintained. There have been more recent releases to adhere to never versions of python. Would you be able to sign up again at https://majiq.biociphers.org/app_download/ and try your installation with the newer version? (python3.6 should work, though python3.8 is recommended) Let me know how this goes for you. 

Thanks, 

-Paul

[Alexander Dietrich]

Hi Paul,

that did the trick! Thank you, I did not realise that I was working with an older version.

Best,

Alex

--
You received this message because you are subscribed to a topic in the Google Groups "majiq_voila" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/majiq_voila/zuOVFNdV3LY/unsubscribe.
To unsubscribe from this group and all its topics, send an email to majiq_voila...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/majiq_voila/628acde3-4dd4-4022-a7ee-840b93abd50an%40googlegroups.com.

[Alexander Dietrich]

Hi Paul,

a follow up question: is it possible to have the A3SS, A5SS and ES columns in the voila output?

Alex

[Paul Jewell]

Hello Alex, 

These column definitions were removed in general, we we ended up not fully trusting them, as we came across more complex splicing patterns, and a simple label per LSV was deemed inappropriate. 

For the next version of majiq, we will have a new separate component which is intended to quantify and more accurately describe these complex splicing events. 

-Paul