Benchmarking Ligand-Based Virtual High-Throughput Screening with the PubChem Database

Abstract

With the rapidly increasing availability of High-Throughput Screening (HTS) data in the public domain, such as the PubChem database, methods for ligand-based computer-aided drug discovery (LB-CADD) have the potential to accelerate and reduce the cost of probe development and drug discovery efforts in academia. We assemble nine data sets from realistic HTS campaigns representing major families of drug target proteins for benchmarking LB-CADD methods. Each data set is public domain through PubChem and carefully collated through confirmation screens validating active compounds. These data sets provide the foundation for benchmarking a new cheminformatics framework BCL::ChemInfo, which is freely available for non-commercial use. Quantitative structure activity relationship (QSAR) models are built using Artificial Neural Networks (ANNs), Support Vector Machines (SVMs), Decision Trees (DTs), and Kohonen networks (KNs). Problem-specific descriptor optimization protocols are assessed including Sequential Feature Forward Selection (SFFS) and various information content measures. Measures of predictive power and confidence are evaluated through cross-validation, and a consensus prediction scheme is tested that combines orthogonal machine learning algorithms into a single predictor. Enrichments ranging from 15 to 101 for a TPR cutoff of 25% are observed.

Improved leaps and bounds variable selection algorithm based on principal component analysis

“…The SFFS algorithm can get a relatively good result in a short time. In some papers, SFFS was used as a variable selection method and achieved good results [27]. The backward methods cannot be used because they have the same problem as the leaps and bounds algorithm that evaluation cannot be done at the first stage.…”

Bioinformatics in Drug Development

“…"DrugBank" is another database which stores data about various drugs and their properties and associates them with target's data; their "sequence, structure and pathway" [12]. "PubChem" is another useful database containing of chemical compounds and is maintained by NCBI [13].…”

PubChem applications in drug discovery: a bibliometric analysis

“…The abundant structure and bioassay data in PubChem have motivated the development and validation of various informatics methods and algorithms [26–28]. For instance, Feldman et al applied PubChem structures to a chemical ontology algorithm that was developed based on chemical functional groups [26].…”

“…Matlock et al demonstrated their scaffold discovery framework by employing the PubChem HTS datasets to maximize the active scaffolds that increased the number of active molecules confirmed by experiments [27]. Butkiewicz and co-workers assembled a group of PubChem HTS datasets for benchmarking ligand-based virtual screening involving the major families of drug target proteins [28]. …”