QSTR with Extended Topochemical Atom Indices. 7. QSAR of Substituted Benzenes toSaccharomyces cerevisiae
Kunal Roy, Indrani Sanyal · QSAR & Combinatorial Science · 2006
Abstract The experimental determination of toxicological properties of commercial chemicals being costly and time consuming process, there is the need to develop mathematical predictive tools to theoretically quantify such properties. In this background, we have modeled the nonspecific toxicity of 51 substituted benzenes to the yeastSaccharomyces cerevisiaeusing extended topochemical atom (ETA) indices. Principal component factor analysis (FA) was used as the data‐preprocessing step to reduce the dimensionality of the data matrix and identify the important variables that are devoid of collinearities. Multiple linear regression (MLR) analyses show that the best ETA model has the following statistical quality:n= 51,Q2= 0.851,Ra2= 0.874,R= 0.940,F= 87.9 (df4, 46),s= 0.235,PRESS= 3.3. We have also modeled the toxicity data using other topological descriptors including Wiener, HosoyaZ, molecular connectivity, kappa shape, BalabanJand E‐state parameters apart from physicochemical parameters like AlogP98, MolRef, H_bond_acceptor and H_bond _donor. The best model shows the following quality:n= 51,Q2=0.837,Ra2 = 0.855,R= 0.929 ,F= 98.9 (df3,47),s= 0.253,PRESS=3.6. An attempt to use a combined set including both ETA and non‐ETA parameters comes out with the following results:n=51,Q2=0.824,Ra2=0.852,R= 0.940,F= 73.0 (df4,46),s= 0.255,PRESS=3.9. Besides FA‐MLR, stepwise regression analysis and partial least squares (PLS) analysis were used as additional statistical tools. The use of the ETA indices suggested negative contributions of functionalities of amino and carboxylic acid substituents on the benzene ring and the presence of the electronegative atoms and positive contributions of branching and functionality of chloro substituent. Using factor scores as independent variables, principal component regression analysis (PCRA) was performed and the derived relations were of the following statistical qualities:Q2values being 0.926, 0.878 and 0.869 whileR2values being 0.942, 0.903 and 0.899 for factor scores derived from ETA, non‐ETA and combined matrices respectively. Thus, it appears that the ETA descriptors have significant potential in QSAR/QSPR/QSTR, which warrants extensive evaluation.