Novel approach to evolutionary neural network based descriptor selection and QSAR model development

Abstract

Capability of evolutionary neural network (ENN) based QSAR approach to direct the descriptor selection process towards stable descriptor subset (DS) composition characterized by acceptable generalization, as well as the influence of description stability on QSAR model interpretation have been examined. In order to analyze the DS stability and QSAR model generalization properties multiple random dataset partitions into training and test set were made. Acceptability criteria proposed by Golbraikh et al. [J. Comput.-Aided Mol. Des., 17 (2003) 241] have been chosen for selection of highly predictive QSAR models from a set of all models produced by ENN for each dataset splitting. All QSAR models that pass Golbraikh's filter generated by ENN for each dataset partition were collected. Two final DS forming principles were compared. Standard principle is based on selection of descriptors characterized by highest frequencies among all descriptors that appear in the pool [J. Chem. Inf. Comput. Sci., 43 (2003) 949]. Search across the model pool for DS that are stable against multiple dataset subsampling i.e. universal DS solutions is the basis of novel approach. Based on described principles benzodiazepine QSAR has been proposed and evaluated against results reported by others in terms of final DS composition and model predictive performance.

Designer benzodiazepines’ pharmacological effects and potencies: How to find the information

“…BZDs act as a positive allosteric modulator on this receptor that is a ligand-gated chloride-selective ion channel. Structure–activity relationship studies and research using neural artificial networks have predicted the binding affinity of several BZDs and DBZDs to the GABA A receptor (Debeljak et al, 2005; Maddalena and Johnston, 1995; So and Karplus, 1996). Waters et al used a qualitative structure–activity relationship (QSAR) model to predict the GABA A5 receptor binding of DBZDs (Waters et al, 2018).…”

“…Because of GABA A receptor complexity, these studies have only uncovered the tip of the psychopharmacological iceberg. Even though these studies have enabled the prediction of the relative potency of several BZDs, they have failed to evidence a relationship between the structure and the main effects of each individual BZD (Debeljak et al, 2005; Maddalena and Johnston, 1995; So and Karplus, 1996).…”

QSAR analysis of diaryl COX-2 inhibitors: Comparison of feature selection and train-test data selection methods

“…Recently, a number of compounds that occur in nature [1] and are synthetic trans stilbenoid [4,5] were evaluated as COX-2 selective inhibitors and their structure e activity relationships were investigated [6].…”

QSAR Study of Antimicrobial Activity of Some 3-Nitrocoumarins and Related Compounds

“…SAR. Antimicrobial activities (Table 1) reveal that 3-nitro-4-(imidazol-1-yl)coumarin (compound 7), 3-nitro-4-(benzimidazol-1-yl)coumarin (compound 8), 3-amino-4-anilinocoumarin derivatives (compounds [22][23][24][25][26][27][28], and 3-nitro-4-(pyridin-2-ylimino)chromen-2-ol derivatives (compounds 29-33) represent broadly potent molecules. The presence of a quaternary ammonium group (compounds 2-6) does not seem to make either a negative or a positive contribution to antimicrobial activity against any of the analyzed microorganisms.…”

“…However, these loops were strictly separated, and the external LOO validation realized this way represents a reliable model evaluation tool. 26,27 There are other external validation protocols available like leave-many-out cross-validation, but a relatively small number of molecules dictates application of LOO. 32 Applied learning methods involve the usage of random number generators as a part of the learning process.…”