Multixem Refinement pipeline for multiple data sets in structure biology

This is still work in progress.

Installation

Install the latest code from this GitHub repository with pip:

pip install git+https://github.com/MartinMalyMM/multixem.git

It is recommended to use a Python virtual environment:

python -m venv .venv
source .venv/bin/activate
pip install git+https://github.com/MartinMalyMM/multixem.git

You will also need Servalcat installed, typically from a recent CCP4 installation which also includes the Monomer Library.

Bootstrap

Subcommand bootstrap: Perform the bootstrap protocol - run multiple refinements in parallel against resampled sub data sets.

Example:

multixem bootstrap 10000 \
	--hklin 1PGJ_data.mtz \
	--model 1PGJ_model.pdb \
    --hklin_free 1PGJ_data.mtz \
	--prefix 1PGJ_bootstrap10000 \
	--servalcat_args "--ncycle 10" \
    --servalcat_confing "config.yaml" \
	--n_bins 30 \
	--n_proc 16 \
    --geometry_cids 1PGJ_geometry_obj.txt

The file given in 1PGJ_geometry_obj.txt defines parameters/features of the structure model under investigation. Each lines specifies an object using atomic CIDs divided by spaces. One CID in row denotes an occupancy, two an interactomic distance, three an angle between the atoms and four a torsion angle. For instance:

//A/505/O2
//A/505/O2 //A/262/CG
//A/505/O2 //A/262/CG //A/262/CD

Regarding the distances and angles, the procedure is suitable for investigation non-covalently linked atoms. The environment is locally unrestrained (Van der Waals anti-bumping restraints switched off) to provide unbiased results.

All currently available options are listed using:

multixem bootstrap --help

Comparison of isomorphous data sets

Subcommand pipeline: Compare the given diffraction data sets, refine given structure model(s) against them and compare them including calculation of isomorphous difference density maps (|Fobs,n|e^iɸn – k|Fobs,1|e^iɸ1). The input data set should be cut at the same resolution.

Example: comparison of bovine, pork and human insulin:

multixem pipeline \
	--hklin insuling_cow.mtz insuling_pig.mtz insuling_people.mtz \
	--hklin_free insuling_people.mtz \
	-p insulin \
	--model insuling_cow.pdb insuling_pig.pdb insuling_people.pdb \
	--n_bins 30 \
	--n_proc 4 \
	--unify_cell```

Example: Merging and comparison of batches of unmerged diffraction data:

multixem pipeline \
	--hklin_unmerged insuling_people_unmerged.mtz \
	--hklin_free insuling_people.mtz \
	-p insulin \
	--model insuling_people.pdb \
	--n_bins 30 \
	--n_proc 4 \
    --n_batches 600
	```

All currently available options are listed using:

multixem pipeline --help

Tests

Automatic tests are implemented using pytest. They can be run using the following command:

python -m pytest -s -v

The integration tests use data from 6-phosphogluconate dehydrogenase (PDB 1PGJ, manuscript) and human, porcine, and bovine insulin (manuscript, raw data (DOI 10.5281/zenodo.13890874)).