Chemically Advanced Template Search (CATS) for Scaffold‐Hopping and Prospective Target Prediction for ‘Orphan’ Molecules

Reutlinger, Michael; Koch, Christian P.; Reker, Daniel; Todoroff, Nickolay; Schneider, Petra; Rodrigues, Tiago; Schneider, Gisbert

doi:10.1002/minf.201200141

Cited by 145 publications

(175 citation statements)

References 56 publications

Supporting

Mentioning

149

Contrasting

Unclassified

Order By: Relevance

“…Chemically abstract ("fuzzy") molecular representations, such as pharmacophoric feature descriptors, can be used to find subtle functional relationships between compounds, thereby allowing a molecule to leapfrog onto an unrelated target (28,29). When used in similarity searches, such fuzzy molecular representations have often demonstrated greater scaffold-hopping potential than atomistic approaches (10,30). Consequently, we implemented SPiDER as a software tool that builds on fuzzy molecular representations for use with de novo-designed NCEs.…”

Section: Resultsmentioning

confidence: 99%

“…Using the unsupervised SOM algorithm, we clustered 12,661 manually annotated, pharmaceutically relevant drugs and lead compounds [collection of bioactive reference analogs (COBRA); inSili.com LLC] (33). The resulting 2D map tessellates this reference space into clusters of drug molecules representing local neighborhoods (10,(34)(35)(36). A query compound is assigned to exactly one target cluster on this map.…”

Section: Resultsmentioning

confidence: 99%

“…Averaging the false-positive error probabilities for the reference drugs annotated to bind to different targets emulates a false-discovery rate in multiple hypothesis testing and provides a motivated confidence score for the prediction. SPiDER performs the target prediction twice, using the following two different molecular representations for two SOM projections: (i) the chemically advanced template search [chemically advanced template search (CATS)] topological pharmacophore descriptor (SOM1) (10,37) and (ii) the Molecular Operating Environment (MOE) physicochemical properties and indices (SOM2) (Chemical Computing Group). Jury consensus predictions are obtained as the average of the two confidence scores.…”

Section: Resultsmentioning

confidence: 99%

“…The 210-dimensional CATS vectors were calculated with an in-house software tool with a maximal correlation distance of 10 bonds and pharmacophoic feature type-sensitive descriptor scaling (10). The following 186 descriptors from MOE were calculated using the MOE "QSAR descriptors" function (MOE 2011.10): simple descriptors, chi indices, BCUT descriptors, PEOE descriptors, Q_charge descriptors, Lipinski rule of fives, Kier indices, GCUT descriptors, SlogP descriptors, VSA descriptors, SMR descriptors, and "other 2D descriptors."…”

Section: Data Preparation and Molecular Representationmentioning

confidence: 99%

“…We report the development, implementation, and successful prospective application of an innovative computational technique for the target profiling of de novo-designed molecules. The approach combines the concepts of self-organizing maps (SOMs) (9) for macromolecular target prediction (10), consensus scoring (11), and a statistical evaluation that provides confidence estimates for the predictions.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Identifying the macromolecular targets of de novo-designed chemical entities through self-organizing map consensus

Reker

Rodrigues

Schneider

2014

Proc. Natl. Acad. Sci. U.S.A.

Self Cite

205

165

View full text Add to dashboard Cite

De novo molecular design and in silico prediction of polypharmacological profiles are emerging research topics that will profoundly affect the future of drug discovery and chemical biology. The goal is to identify the macromolecular targets of new chemical agents. Although several computational tools for predicting such targets are publicly available, none of these methods was explicitly designed to predict target engagement by de novo-designed molecules. Here we present the development and practical application of a unique technique, self-organizing map-based prediction of drug equivalence relationships (SPiDER), that merges the concepts of self-organizing maps, consensus scoring, and statistical analysis to successfully identify targets for both known drugs and computer-generated molecular scaffolds. We discovered a potential off-target liability of fenofibraterelated compounds, and in a comprehensive prospective application, we identified a multitarget-modulating profile of de novo designed molecules. These results demonstrate that SPiDER may be used to identify innovative compounds in chemical biology and in the early stages of drug discovery, and help investigate the potential side effects of drugs and their repurposing options.drug design | target prediction | polypharmacology | machine learning | chemical similarity C omputer-assisted de novo molecular design has evolved as a popular source of ideas to combat the perceived lack of new chemical entities (NCEs) in chemical biology and drug discovery (1). We demonstrate that automated de novo design delivers readily synthesizable NCEs with desirable activity profiles. Although receptor-based design operates on a model of a macromolecular binding site, ligand-based methods are either explicitly or implicitly based on the chemical similarity principle (2) without requiring a receptor model (3). Instead, the latter typically uses some measure of chemical or pharmacophore feature similarity to a reference ligand as a fitness function, which aims to generate NCEs as template mimetics via scaffold hopping (4-8). We report the development, implementation, and successful prospective application of an innovative computational technique for the target profiling of de novo-designed molecules. The approach combines the concepts of self-organizing maps (SOMs) (9) for macromolecular target prediction (10), consensus scoring (11), and a statistical evaluation that provides confidence estimates for the predictions.Predicting polypharmacological activities is a topic relevant to chemical biology and drug discovery, not only to take advantage of inherent drug promiscuity but also to decrease lead compound attrition caused by unfavorable off-target modulation

show abstract