| title | tags | authors | affiliations | date | bibliography | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
kallisto: A command-line interface to simplify computational modelling and the generation of atomic features |
|
|
|
8 March 2021 |
paper.bib |
Machine learning (ML) has recently become very popular within pharmaceutical industry [@roy2015; @sprous2010]. Tasks as, e.g., building predictive models, performing virtual screening, or predicting compound activities are potential use cases for such ML applications [@li2021; @simm2018]. Traditionally, ML models often rely on the quantitative structure-activity relationship (QSAR) that has been popularized by medicinal chemists and statisticians to relate bioactivities to specific functional group manipulations [@dudek2006; @verma2010]. This QSAR approach decreases the dimensionality of the underlying problem and projects the molecular structure into a space spanned by the physicochemical features. While early approaches relied more on linear regression, modern approaches combine such features with non-linear ML algorithms.
Chemoinformatic packages like RDKit [@landrum2006] enable the fast calculation of atomic/molecular features based on structural information like the molecular graph, while recently an extended Hueckel package has been added as well [@landrum2019]. However, frequently we want to go beyond a structure-only approach thus incorporating electronic structure effects as obtained, e.g., by a (higher-level) quantum mechanical (QM) treatment. The calculation of QM-based features relies often on well-established quantum chemistry methods like Kohn-Sham density functional theory (DFT) that is currently the workhorse of computational chemistry [@parr1980; @kohn1999]. However, generating the feature space by DFT is computationally demanding and can become the computational bottleneck especially when aiming for high-throughput experiments with several hundred to thousands of molecules.
Since there exists a critical need for an efficient yet accurate featurizer, we developed the kallisto command-line interface that is able to calculate QM-based atomic features for atoms and molecules efficiently (whole periodic table up to Radon).
The features are either interpolating high-level references (e.g., static/dynamic polarizabilities with time-dependent DFT data) or are parametrized [@caldeweyher2019] to reproduce QM references (e.g., DFT Hirshfeld [@hirshfeld1977] atomic partial charges).
Molecular geometries need to have an xmol or a Turbomole like format to be processed by kallisto.
Besides, we implemented several computational modelling helpers to simplify the development of high-throughput procedures.
Some of those modelling helpers depend on the open-source xtb tight-binding scheme that has been developed by Stefan Grimme and co-worker [@bannwarth2020].
The kallisto software depends on the scientific libraries Numpy [@harris2020] and SciPy [@virtanen2020].
The online documentation covers all high-level functionalizations of this software mostly in terms of copy-paste recipes.
Furthermore, we cover bits of the underlying theory and compare to experimental data as well as to other modern deep learning models.
The following atomic and molecular features are available for all atoms up to Radon
- Coordination numbers [@grimme2010; @caldeweyher2019]
- Proximity shells
- Environment-dependent electronegativity equilibration partial charges [@caldeweyher2019]
- Environment- and charge-dependent dynamic polarizabilities [@grimme2010; @caldeweyher2019]
- Environment- and charge-dependent van-der-Waals radii [@fedorov2018; @rahm2017; @mantina2009]
- Sterimol descriptors (L, Bmin, Bmax) [@brethome2019]
The following modelling helper are implemented
- Breadth first sorting
- Root mean squared deviation (quaternions) [@coutsias2004]
- Substructure identifier
- Substructure exchanger
EC acknowledges contributions from Philipp Pracht (@pprcht) and thanks Kjell Jorner (@kjelljorner) for sharing his Sterimol algorithm.