pyCOSMOS: python for Compartmentalization Of Solid Metal-Organic framework Structures

pyCOSMOS partitions the unit cell of a MOF into distinct pore compartments. The unit cell is subdivided into cubelets, which are then classified into a pore types. For instance, PCN-224 comprises two pore types. Through the use of pyCOSMOS, the cubelets within the unit cell are designated as green or yellow, representing the channel and interaction pores, respectively.

Requirements

1. cython: pip install cython
2. mayavi (for visualization): pip install mayavi
3. openbabel (for converting cif to pdb): sudo snap install openbabel
4. Python libraries: pandas, numpy, plotly, sklearn, matplotlib, os, glob.

How to use?

Build the cython file:

python3 setup_periodic_distance.py build_ext --inplace It will generate a folder build and the file periodic_distance.cpython-311-x86_64-linux-gnu.so

Required: Perform Zeo++ pore size distribution calculation

1. Install Zeo++ from: http://zeoplusplus.org/. After successful installation, network executable is generated.
2. Perform psd calculation using: ./network -ha -vpsd 1.657 1.657 50000 Structure.cif. This will generate *vpsdpts files. Here 1.657 is the probe radius and 50,000 monte carlo insertions are attempted to calculate the pore size distribution.

Inputs

input.txt

vpsdpts  file.vpsdpts
lx       38.8050
ly       38.8050
lz       38.8050
alpha    90.0000
beta     90.0000
gamma    90.0000
npore    2
eps      2.500
nmin     25

1. vpsdpts: pore size distribution file from Zeo++
2. npore: Number of pore types. (If you don't know this, run this code with guess value of 1, 2, and 3 in this order)
3. lx, ly, lz, alpha, beta, gamma are unit cell dimensions.
4. eps and nmin are the parameters of the DBSCAN clustering algorithm.

To run pycosmos:

python pyCOSMOS/src/main.py input.txt

Outputs

The results of the algorithm is in the form of a pore type matrix of size $$(lx \times ly \times ly)$$.

csv files:

1. pore_type_matrix_with_cluster_labels.csv
2. pore_type_matrix_with_pore_type_labels.csv

html file (can be viewed on browser)-

1. pore_type_matrix_with_cluster_center_labels.html
2. pore_type_matrix_with_pore_type_labels.html
3. geometric_points_with_cluster_labels_for_pore_type_*Npores.html

xyz files: for each primary bin in the pore size distribution. Additional xyz files for each cluster within a bin.

png files: characteristic snapshots of the framework with pores visualized as cages or channels.

References

1. Pore Structure Compartmentalization for Advanced Characterization of Metal–Organic Framework Materials
   Shivam Parashar and Alexander V. Neimark
   Journal of Chemical Information and Modeling
   DOI: 10.1021/acs.jcim.3c01872