Cannot source ccp4 and phenix: samefilename issue running matchmaps

mjmcleod64 commented 7 months ago

matchmaps version: install from pip
Python version: 3.9.16
Operating System: ubuntu 22.04

Description

I am pretty unversed in linux in conda installations. I am trying to create my first difference maps after installation. Receving program with sourcing ccp4 and phenix as well as running program.

What I Did

I did the installation as specific conda create -n my-matchmaps-env python=3.9 conda activate my-matchmaps-env pip install matchmaps

source /usr/local/phenix-1.21rc1-5170/phenix_env.sh /opt/xtal/ccp4-8.0/start

This error shows: This script is to be run in subshell, not sourced.

I continued forward and attempted to run the program using: matchmaps --mtzoff 2qew_phases.mtz FP SIGFP --mtzon 3dt2_phases.mtz FP SIGFP --pdboff 2qew.pdb where the files are present in the directory I am in

I receive this error: Traceback (most recent call last): File "/home/mjmcleod/eman2-sphire-sparx/envs/my-matchmaps-env/bin/matchmaps", line 8, in sys.exit(main()) File "/home/mjmcleod/eman2-sphire-sparx/envs/my-matchmaps-env/lib/python3.9/site-packages/matchmaps/_compute_realspace_diff.py", line 450, in main compute_realspace_difference_map( File "/home/mjmcleod/eman2-sphire-sparx/envs/my-matchmaps-env/lib/python3.9/site-packages/matchmaps/_compute_realspace_diff.py", line 109, in compute_realspace_difference_map pdboff = _cif_or_pdb_to_pdb(pdboff, output_dir) File "/home/mjmcleod/eman2-sphire-sparx/envs/my-matchmaps-env/lib/python3.9/site-packages/matchmaps/_utils.py", line 979, in _cif_or_pdb_to_pdb shutil.copy(input_file, output_file) File "/home/mjmcleod/eman2-sphire-sparx/envs/my-matchmaps-env/lib/python3.9/shutil.py", line 427, in copy copyfile(src, dst, follow_symlinks=follow_symlinks) File "/home/mjmcleod/eman2-sphire-sparx/envs/my-matchmaps-env/lib/python3.9/shutil.py", line 244, in copyfile raise SameFileError("{!r} and {!r} are the same file".format(src, dst)) shutil.SameFileError: PosixPath('2qew.pdb') and PosixPath('2qew.pdb') are the same file

And am now stuck. Any help would be appreciate

dennisbrookner commented 7 months ago

Hey, thanks so much for reaching out and for raising this error!

First, the quick fix: this error should be resolved by adding the --output-dir option to your matchmaps call, e.g.:

matchmaps --mtzoff 2qew_phases.mtz FP SIGFP --mtzon 3dt2_phases.mtz FP SIGFP --pdboff 2qew.pdb --output-dir matchmaps_files

dennisbrookner commented 7 months ago

Source of this error

matchmaps is designed to work as much as possible in the output directory, and interfere minimally with the input directory. As such, early in the program, various input files are copied to the output directory. However, this behavior does not account for the possibility that the input and output directories are the same!

Specifically, if it is true that both:

The input and output directories are identical (presumably because they are both ./)
The starting model is provided in .pdb format

then the above error will always occur.

Regretfully, as this situation was not foreseen, the error message above is quite unhelpful.

Potential solutions

1. Make `shutil.copy` call ignore the error

Replacing

shutil.copy(input_file, output_file)

with

try:
    shutil.copy(input_file, output_file)
except shutil.SameFileError:
    pass

in line 1016 of _utils.py would solve this issue.

It would be important to further test this solution to ensure that it does not cause unexpected errors downstream!

2. If `--output-dir` is not provided, use a default value other than `./`

I don't really like this option, because it has the potential to create an unexpected directory structure for the user. I would also have to handle the case where a directory with the default name already exists; do you use it then? Or do you slap a _1 on the end?

3. Make `--output-dir` a required parameter

The downside of this approach is that adding another required parameter makes matchmaps calls necessarily longer and clunkier. The advantage is that the user maintains full control over the resulting directory structure. Outputs could still be directed to the current directory by supplying --output-dir ./, for example.

Interplay between these options

Change 3 would require making change 1 as well; otherwise, supplying --output-dir ./ would just recreate the original error! If change 2 were made, change 1 should not be made. Instead, the SameFileError should be raised with an informative error message that explains what has happened and how to fix it.

Conclusion

I will consider the above options and push an appropriate fix in the immediate future.

dennisbrookner commented 7 months ago

source /usr/local/phenix-1.21rc1-5170/phenix_env.sh /opt/xtal/ccp4-8.0/start

This error shows: This script is to be run in subshell, not sourced.

I just realized that I didn't respond to this part. Confusingly, the phenix environment is activated with the source command:

source /usr/local/phenix-1.21rc1-5170/phenix_env.sh

whereas ccp4 comes with its own executable that should just be run as:

/opt/xtal/ccp4-8.0/start

with no source command out front.

Given the subsequent error you described above, which comes later in the program, it seems that you were able to get CCP4 sourced successfully. In any case, I'm adding this here for the sake of completeness.

dennisbrookner commented 7 months ago

This bug is fixed as of PyPI version 0.6.2. Now, when the input and output directories match, a warning message is printed out to the user, and the program proceeds.

rs-station / matchmaps