Closed MikeWLloyd closed 2 years ago
Hi, -d option is not required in msisensor2. So, just run:
/home/lloydm/msisensor2/msisensor2 msi -M /home/lloydm/msisensor2/models_hg38 -t
I will update readme soon.
Will running with -d affect the final metrics? Can you expand on why it is not required?
Also, running without the -d flag as you suggest did not fix the issue. Values still return as zeros. When the user copies the models directory to their local directory, and runs the tool, the run finishes with expected values.
Owing to the machine learning method, the distinguishing sites are selected, so the -d parameter is no longer needed to specify the bed file.
If the expected results can be obtained by changing the directory. Then you need to verify:
Will running with -d affect the final metrics? Can you expand on why it is not required?
Also, running without the -d flag as you suggest did not fix the issue. Values still return as zeros. When the user copies the models directory to their local directory, and runs the tool, the run finishes with expected values.
Yes, I have the same problem. I set the write permissions to folders for other users and it work. I think it's dangerous if someone changed the files。
I am working to implement your tool on a compute cluster. I am having an issue, and I wonder if you could provide input.
When I run with local files:
The tool provides valid output:
When a different user tries to run the same command with the same inputs, the tool returns this output with no other apparent differences in the output log.
I can only assume there are permissions issues involved here, but in my view, the tool should not produce output with zeros, if it failed to write to a location or load a particular file needed.