cell2mol: encoding chemistry to interpret crystallographic data

The creation and maintenance of crystallographic data repositories is one of the greatest data-related achievements in chemistry. Platforms such as the Cambridge Structural Database host what is likely the most diverse collection of synthesizable molecules. If properly mined, they could be the basis for the large-scale exploration of new regions of the chemical space using quantum chemistry (QC). However, it is currently challenging to retrieve all the necessary information for QC based exclusively on the available structural data, especially for transition metal complexes. To solve this shortcoming, we present cell2mol, a software that interprets crystallographic data and retrieves the connectivity and total charge of molecules, including the oxidation state (OS) of metal atoms. We prove that cell2mol outperforms other popular methods at assigning the metal OS, while offering a much more comprehensive interpretation of the unit cell, and we make publicly available reliable QC-ready databases totaling 31k transition metal complexes and 13k ligands, encompassing incomparable chemical diversity. This record contains the aforementioned database of crystallographic structures after interpretation using the cell2mol software. The database spans 8 different transition metals (Fe, Mn, Ru, Re, Cr, Co, Ni, Cu; named from 1 to 8) and contains over 31000 different transition metal complexes and 13000 unique ligands, but also contains the interpreted contents of the entire unit cells in terms of discrete chemical species with well-defined charges and connectivities. Details can be found in the README.txt file and an exemplary script is provided for usage. The cell2mol code can be obtained in https://github.com/lcmd-epfl/cell2mol.