Free-Wilson-Type Analysis of Non-Additive Substituent Effects on THPB Dopamine Receptor Affinity Using Artificial Neural Networks

Abstract

Recently published brain dopamine D 2 receptor af®nity data of 15 tetrahydroprotoberberine (THPB) derivatives acting as dopamine receptor antagonists have been analyzed by two different QSAR techniques. The following main results were obtained by this analysis: In contrast to an unsuccessful Free-WilsonyFujita-Ban analysis the investigated receptor binding data could be described by a neural network approach using only binary substructural indicator variables. The arti®cialycomputational neural network was able to recognize that the af®nity depends signi®cantly on the simultaneous presence or absence of two or more substituents. A 4-D plot demonstrates the non-additivityyvariability of substituent effects on D 2 receptor af®nity.

Synthesis, Molecular Modeling, and Quantitative Structure–activity Relationship Studies of Undec-10-Enehydrazide Derivatives as Antimicrobial Agents

“…Further, the regression analysis and development of QSAR models was performed using the TSAR 3.3 software. The predictive powers of the developed models were supported by crossvalidated r 2 (q 2 ) using leave one out (LOO) cross-validation method [37]. The statistical qualities of equations were further confirmed by the parameters such as standard error of estimate (s), correlation coefficient (r), variance ratio (F) at specified degrees of freedom.…”

“…The predictive powers of derived QSAR models were confirmed by LOO method [37], where a model is built with N -1 compounds and N th compound is predicted. Each compound is eliminated for model derivation and predicted in turn.…”

QSAR Study of 3-Phenyl-5-acyloxymethyl-2H,5H-furan-2-ones as Antifungal Agents: The Dominant Role of Electronic Parameter

“…27,28) Predicted residual sum of square (PRESS), total sum of squares (SSY), cross-validated R 2 (Q 2 ), and standard deviation error of prediction (SDEP) were considered for the validation of these models.…”

Antimicrobial evaluation of 4-methylsulfanyl benzylidene/3-hydroxy benzylidene hydrazides and QSAR studies

“…Further, the regression analysis was performed using the SPSS software package (SPSS for windows, version 10.05, 1999). The predictive powers of the equations were validated by leave one out (LOO) cross-validation method (Schaper, 1999), where a model is built with N -1 compounds and Nth compound is predicted. Each compound is left out of the model derivation and predicted in turn.…”