Prediction of chemical–protein interactions: multitarget-QSAR versus computational chemogenomic methods

Abstract

Elucidation of chemical-protein interactions (CPI) is the basis of target identification and drug discovery. It is time-consuming and costly to determine CPI experimentally, and computational methods will facilitate the determination of CPI. In this study, two methods, multitarget quantitative structure-activity relationship (mt-QSAR) and computational chemogenomics, were developed for CPI prediction. Two comprehensive data sets were collected from the ChEMBL database for method assessment. One data set consisted of 81 689 CPI pairs among 50 924 compounds and 136 G-protein coupled receptors (GPCRs), while the other one contained 43 965 CPI pairs among 23 376 compounds and 176 kinases. The range of the area under the receiver operating characteristic curve (AUC) for the test sets was 0.95 to 1.0 and 0.82 to 1.0 for 100 GPCR mt-QSAR models and 100 kinase mt-QSAR models, respectively. The AUC of 5-fold cross validation were about 0.92 for both 176 kinases and 136 GPCRs using the chemogenomic method. However, the performance of the chemogenomic method was worse than that of mt-QSAR for the external validation set. Further analysis revealed that there was a high false positive rate for the external validation set when using the chemogenomic method. In addition, we developed a web server named CPI-Predictor, , which is available for free. The methods and tool have potential applications in network pharmacology and drug repositioning.

Drug repositioning and repurposing: terminology and definitions in literature

“…These definitions represent the range of definitions from nonspecific to specific as observed in those 67 articles. For instance Cheng et al referred just to 'new usages' [17] whereas Sistigu et al specifically stated: 'novel indication underscoring a new mode of action that predicts innovative therapeutic options' [18].…”

GPCR & Company: Databases and Servers for GPCRs and Interacting Partners

“…The user inserts the compound of interest (either by structure or in SMILES format) and the server returns target GPCRs predicted to bind this compound. This predictor has the potential to facilitate applications in network pharmacology and drug repositioning (Cheng et al 2012). …”

“…IUPHAR-DB contains curated information on protein superfamilies that are major biological targets of licensed medicinal drugs. This includes almost 360 GPCRs from human and Cheng et al (2012) The data from all these databases can be downloaded. Some parts of the data in GPCR IUPHAR-DB cannot be downloaded but can be given upon request and they are free for academic users.…”

CompoundProtein Interaction Prediction Within Chemogenomics: Theoretical Concepts, Practical Usage, and Future Directions

“…[53] with the PROFEAT protein sequence analysis suite [54] to build random forest and support vector machine (SVM) models for CPI prediction. [48] Cheng et al combined the 166 fixed-substructure MACCS key method with the PROFEAT method, and tested a large collection of single protein SVM models (multitarget QSAR) in comparison to a chemogenomics-type approach, [50] and concluded with caution on being over-confident of model performance based only on standard crossvalidation of chemogenomic models. In a completely different method, Keiser and colleagues organized several hundred receptors into a hierarchy based only on the similarity of each receptor's known ligands (no protein information), with the emergence of biogically-relevant clusters and several novel ligand-target interactions confirmed experimentally.…”