Quantitative descriptions of molecules are necessary for machine learning algorithms or multivariate linear regressions in order to construct a model to predict reaction outcomes such as percent yield, selectivity, or other experimental measures of reactivity. Looking into quantifiable methods to describe molecular shape, it can be noted that current measures are limited in what aspects of steric behavior they are able to describe, often represented as a single value. The Sterimol parameter set is made up of three subparameters describing the length of a molecule and minimum and maximum widths along the length. These commonly used steric descriptors rely on arbitrarily defined van der Waals radii that fail to capture nuances of molecular shape. Current steric parameters can be improved by robustly describing molecular contour. By computing Sterimol minimum and maximum widths iteratively along the length of a molecule in slices, we are given insight on the proximal and distal steric effects of a substituent. These scans present a new way to embed molecular structure and shape, giving quantitative ways to describe molecular shape with more dimensionality as molecular descriptors.<br/><br/>We have developed a Python package to compute these Sterimol scan parameters, implemented for use on the command line or as an importable module for workflow use. Input options include isodensity surface files, which uses electron density to calculate these parameters, however, standard XYZ coordinate files may also be used, in which case van der Waals radii are employed. This program also allows for the calculation of additional steric parameters, such as percent buried volume. Results are printed out for the user or stored in a Python object for later retrieval. This code will be made freely available and easily downloaded from our research group GitHub.

Sterimol2Vec: Adding new dimensions to sterimol parameters

ACS Fall 2020 Virtual Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.