Table 2.
Examples of in silico programs for SOM prediction.
| Program/reference | Description | Homepage |
|---|---|---|
| Target-based methods | ||
| Tarcsay et al. (2010) | SOM selection is based on docking and binding energies of substrates’ metabolites. | – |
| Vasanthanathan et al. (2009) | Active conformations of CYP1A2 substrates are recognized by docking and binding energy calculation. | – |
| Ligand-based methods | ||
| META-PC | Predicts the structure of likely metabolites; uses a genetic algorithm to prioritize a large biotransformations dictionary; uses also QC descriptors. | multicase.com/meta-pc |
| MetabolExpert | Predicts the structures of likely metabolites using a database containing rules including substrate and metabolite listings; also contain lists of substructures which inhibit or promote the reaction | compudrug.com/metabolexpert |
| Meteor Nexus | Knowledge-based software; integrated to SMARTCyp. | lhasalimited.org/products/meteor-nexus.htm |
| MetaPrint2D (Boyer and Zamora, 2002; Boyer et al., 2007; Adams, 2010) | A data-mining tool that identifies SOMs based on circular fingerprints and fragment-based substrate-metabolite occurrence ratios. | www-metaprint2d.ch.cam.ac.uk/ |
| RS-WebPredictor (Zaretzki et al., 2011, 2012) | Generates pathway-independent, CYP form-specific regioselectivity. Models built with machine learning techniques using numerous QC and topological descriptors. | reccr.chem.rpi.edu/Software/RS-WebPredictor/ |
| SMARTCyp (Rydberg et al., 2010, 2013; Rydberg and Olsen, 2011, 2012) | SOM prediction tool that utilizes fragment-based reactivity and accessibility factors. | farma.ku.dk/smartcyp/ |
| XenoSite (Zaretzki et al., 2013) | Uses both atomic and molecular descriptors in CYP form-specific models built with machine learning methods. | http://swami.wustl.edu/xenosite/ |
| Tyzack et al., 2014 | Form-specific machine learning models that use only 2D topological fingerprints as descriptors. | – |
| Combined methods | ||
| MetaSite (Zamora et al., 2003; Cruciani et al., 2005, 2013, 2014) | Identifies likely SOMs by considering reactivity and complementarity of substrate and CYP catalytic site 3D fingerprints; not training set dependent. | moldiscovery.com/software/metasite/ |
| Tyzack et al. (2013) | Utilizes tethered docking, QC activation energies and molecular dynamics. | – |
| DR-Predictor (Huang et al., 2013) | Combines docking-derived binding energies to atomic descriptors in CYP form-specific models built with machine learning methods. | – |
| StarDrop P450 | Uses AM1 hydrogen atom transfer energy calculations combined with accessibility descriptors. | optibrium.com/stardrop/stardrop-p450-models.php |
| IMPACTS (Campagna-Slater et al., 2012) | Combination of docking, transition state modeling, and rule-based substrate reactivity prediction. | http://molecularforecaster.com/products.html fitted.ca/impacts.html |