filter_los_csd

This is the manual for the script which filters line-of-sight contacts from the results of ConQuest search over the Cambridge Structure Database (CSD) entries. The script is available via GitHub. For the underlying scientific context see the corresponding work: DOI 10.1002/cphc.201901083.

If you use the script in your work, please cite generously!

I. Yu. Chernyshov, I. V. Ananyev, E.A. Pidko, Revisiting van der Waals Radii: From Comprehensive Structural Analysis to Knowledge‐Based Classification of Interatomic Contacts. ChemPhysChem 2020, 21, 370–376, DOI: 10.1002/cphc.201901083.

Intro

This manual consists of two parts, describing installation details and syntax of the script. The example of input and output files for the script is provided in example directory.

Installation

filter_los_csd is a multiplatform Python3 package. Use pip to install it:

> pip install filter_los_csd

filter_los_csd depends on the PyCifRW package, which requires a C/C++ compiler, and pip installation will fail if you do not have one. The best choice for Windows is Visual Studio Building Tools, and for Linux, gcc would be enough. If you forget about it, pip will give you the installation error and platform-specific advice on fixing it.

A Permission denied error may occur when running the script in some versions of Linux:

> filter_los_csd.py input.csv input.cif
-bash: /path/to/script/filter_los_csd.py: Permission denied

To fix it, set umask to 022 before script installation (see ref for more details), or just change permissions:

> sudo chmod 755 /path/to/script/filter_los_csd.py

Syntax

Input and Output

The script takes two necessary parameters as input:

• path_csv: path to CSV file, containing info on parameters of contacts in question. Please note, that CSV file must contain labels of atoms forming the contact and the contact distance!
• path_cif: path to multiple CIF file, containing CSD entries from CSV file. Please note, that CIF file must contain info on bonds. Make sure that the corresponding checkbox was selected before downloading CIF. Also note, that script were tested only for CIF files extracted from CSD (ConQuest)!

Thus, the easiest command is:

> filter_csd_los.py test.csv test.cif

As output the script creates {csv_name}_los.csv file containing the same info as the original file with three additional columns:

• "LOS": has four possible values:
  • "+": the corresponding contact is line-of-sight;
  • "–": the corresponding contact is not line-of-sight;
  • "?": the corresponding crystal is larger than specified cell volume cutoff;
  • "!": the error occurred during the calculation. The only type of errors caught during the testing is inability to find a contact with given atomic labels and distance. The main source of these errors are disorder issues, thus tuning --tol parameter can solve the problem.
• "SHIELDING": contact shielding value (the definition of the term is given in the manuscript);
• "SHIELD_ATOM": label of the shielding atom.

The contact and the shielding atom can be visualized in two steps:

1. find the contact by atomic labels and distance;
2. the nearest atom to the contact line with SHIELD_ATOM label is a shielding atom.

Optional parameters

In addition, the script has several optional parameters:

• -r or --radii: type of van der Waals radii used for shielding calculation, "csd" by default. Available values are:
  • "csd": version usually used in CCDC products (ConQuest, Mercury, etc). It is same as Bondi's version but with r(H) = 1.09 Å; unknown atomic radii was set to 2.0 Å;
  • "bondi": Bondi version; unknown atomic radii set to 2.0 Å;
  • "rt": Rowland&Taylor version, unknown atomic radii set to "csd" values;
  • "alv": Alvarez version;
  • "chap": Chernyshov&Ananyev&Pidko (this work) version.
• --lab1: name of CSV file column containing labels of the contact's first atom, "LAB1" by default;
• --lab2: name of CSV file column containing labels of the contact's second atom, "LAB2" by default;
• --dist: name of CSV file column containing contact distances, "DIST1" by default;

Please make sure, that --lab1, --lab2 and --dist values correspond to those in CSV file.

• --norm: type of X–H bonds normalization, "csd" by default. Please make sure, that the same normalization scheme was used for a ConQuest search. Available values are:
  • "csd": C–H, N–H and O–H bonds are normalized to 1.089 Å, 1.015 Å and 0.993 Å, correspondingly;
  • "no": X–H bonds are not normalized;
  • path to the file, each line of those contains space separated element symbol and the length of corresponding X–H bond in angstroms, e.g. "C 1.09".
• --tol: minimal possible distance between different atoms, default 0.005 Å.
• -V or --volume: maximal allowed cell volume, ų. Available values are:
  • positive numeric value: in this case the script does not calculate contact shielding for crystals with crystallographic cell volume more than specified. It can be useful if there are a lot of crystals with large crystallographic cells (V > 10 000 ų) which are treated slowly.
  • unspecified: no filtering by volume is applied.