Closed RickGelhausen closed 4 years ago
@RickGelhausen Thank you for the report. As you pointed at there is no reason for multiple arguments in the predict function of DeepRibo. I have no idea how this has not resulted in any errors for me before. It is fixed now.
I have cleaned up the sys.exit() arguments. They must have been an artifact from when I started creating these files that I have never really thought about.
I hope everything works as expected.
EDIT: I have retagged v1.1 to include the fix.
Thank you for fixing this so quickly and for creating this useful tool.
Hi @jdcla,
thank you again for tagging the new release. I downloaded it and installed the dependencies using your requirements.yaml. When I tested it on some test dataset, I ran into a minor problem, which was probably caused by the updated pandas dependency.
In the CustomLoader, within DeepRibo.py, I encounter the following error message:
when running the following command (the values for rpkm and coverage comming from the
s_curve_cutoff_estimation.R
script):I was able to pinpoint this to an issue that the coverage is somehow saved as a list, rather than a single value, this crashes:
where a value is expected but a list is given.
Is there any reason why the argparse
nargs=+
option is used to store the coverage parameter? As this causes it to be stored as a list?When I change this locally, the prediction works.
Another very small thing. Is there any reason why deepribo is called using a
sys.exit()
command? As this is an error call.This does not change anything when you run the code locally, as the results are written to file and then deepribo "crashes". When using a workflow management system however, this will cause deepribo to crash the entire workflow when it finishes the prediction/training call.
If there is no reason that I am overlooking, you could simply use:
Thank you for your help.