The 40th Anniversary of the Institute of Physiologically Active Compounds of the Russian Academy of Sciences
QSAR Modeling of Acute Neurotoxicity of Some Organic Solvents with Respect to Rodents
Institute of Physiologically Active Compounds of the Russian Academy of Sciences, 1 Severny proezd, Moscow region, Chernogolovka, 142432 Russia;^{*}email: beng@ipac.ac.ru
Key words: acute neurotoxicity; QSAR; HYBOT
DOI: 10.18097/BMCRM00019
Using literature data analysis, the regression models of acute sublethal neurotoxicity of 47 organic solvents with respect to rats and mice have been developed. To construct the models, we used linear regression, random forest and support vector machines approaches. The linear regression equations were selected as the best models. They are designed on the basis of four molecular descriptors: polarizability, sum of positive atom charges, sum of proton acceptor descriptors and dipole moment. The developed models have good descriptive and predictive ability and clear physicochemical interpretation.
Nhe structure of the compounds was described using fifteen physicochemical descriptors to describe including polarizability, partial atomic charges, hydrogen bond descriptors calculated on the basis of the HYBOT program [10], lipophilicity [11] and dipole moment [12]. The following statistical methods were used to create the regression models: linear regression (LR), random forest (RF), support vector machine (SVM) realized in the corresponding computer programs: SVD [13], rrforest [14] and flssvm [15]. As the statistical characteristics of the models the next parameters were taken: n  number of molecules, m – number of descriptors, r^{2} – square of the linear correlation coefficient, s – standard deviation, q^{2} and s_{cv} – square of the linear correlation coefficient and standard deviation in condition of crossvalidation for kblocks (k=10, 100 iterations). Standard errors were used to estimate the error of the LR model coefficients. Regression models were calculated by examinations of all possible combinations of 15 descriptors. The best models were determined by the minimum value of s. In the case of close values of s, models with smaller values of m were preferred. A simple approach was used to determine the applicability domain (AD) of models, which consisted in calculation of the intervals of changing of dependent and independent variables.
CLOSE

Table 1.
Statistical properties of neurotoxicity models against rats.

CLOSE

Table 2.
Statistical properties of neurotoxicity models against mice.

ACKNOWLEDGEMENTS
The work was carried out within the framework of the state task for 2018 (topic number 009020170020).
SUPPLEMENTARY
Supplementary materials are available at http://dx.doi.org/10.18097/BMCRM00019
REFERENCES
 Mironov, A. N., & Bunatyan, N. D. (2012). Guidelines for conducting preclinical trials of medicinal products. Part one. Moscow: Grif and K, 237–238. ISBN 9785990144712
 Makhaeva, G. F., Radchenko, E. V., Baskin, I. I., Palyulin, V. A., Richardson, R. J., Zefirov, N. S.(2012). Combined QSAR studies of inhibitor properties of Ophosphorylated oximes toward serine esterases involved in neurotoxicity, drug metabolism and Alzheimer's disease. SAR QSAR Environ. Res., 23(78), 627–647. DOI
 Mager, P. P. (1982). Structure neurotoxicity relationships applied to organophosphorus pesticides. Toxicol. Lett., 11(12), 67–71. DOI
 Sanz, P., Flores, I.C., Soriano, T., Repetto, G.,Repetto, M. (1995). In vitro quantitative structureactivity relationship assessment of pyrrole adducts production by γdiketoneforming neurotoxic solvents. Toxicology in Vitro, 9(5), 783–787. DOI
 Lapenna, S., FuartGatnik, M. and Worth A. (2010). Review of QSAR Models and Software Tools for predicting Acute and Chronic Systemic Toxicity. EUR – Scientific and Technical Research series. DOI
 Frantik, E., Hornychova, M. and Horvath, M. (1994). Relative Acute Neurotoxicity of Solvents: Isoeffective Air Concentration of 48 Compounds Evaluated in Rats and Mice. Environ. Res., 66(2), 173185 DOI
 Cronin, M. T. D. (1996). Quantitative structureactivity relationship (QSAR) analysis of the acute sublethal neurotoxicity of solvents. Toxicology in Vitro, 10(2), 103 – 110. DOI
 Estrada, E., Molina, E. and Uriarte, E. (2001). Quantitative StructureToxicity Relationships Using TopsMode. 2. Neurotoxicity of a NonCongeneric Series of Solvents. SAR QSAR Environ. Res., 12(5), 445459. DOI
 Moser, V. C. (2011). Functional Assays for Neurotoxicity Testing. Toxicol. Pathol., 39(1), 36–45. DOI
 Raevsky, O. A., Grigor’ev, V. Yu. , Trepalin, S. V. (1999). HYBOT (Hydrogen Bond Thermodynamics). Russian State Patent Department (registration number № 990090).
 Raevskii, O. A., Trepalina, E. P. & Trepalin, S. V. (2000). Slipper: A new program for calculating solubility, lipophilicity, and liposome penetrability. Pharm. Chem. J., 34(1), 3336. DOI
 Osipov, O. A., Minkin, V. I., & Garnovskii, A. D. (1971). Handbook of Dipole Moments. Vysshaya Shkola, Moscow, 180.
 Forsythe, J., Malcolm, M., & Mowler, C. Computer Methods of Mathematical Calculations [Russian translation], Mir, Moscow (1980). Google Scholar, 210.
 Program RRforest. Retrieved May 05, 2018 from http://www.stat.berkeley.edu/~breiman/RandomForests/reg_examples/RFR.f
 Support Vector Machine aplication written in fortran 90. Retrieved May 05, 2018 from https://github.com/jbcolme/fortranlssvm
 OECD principles for the validation, for regulatory purposes, of (quantitative) structureactivity relationship models. Retrieved May 05, 2018 from http://www.oecd.org/dataoecd/33/37/37849783.pdf
 Tropsha, A., Gramatica, P. and Gombar, V. K. (2003). The Importance of Being Earnest: Validation is the Absolute Essential for Successful Application and Interpretation of QSPR Models. QSAR Comb. Sci., 21(1), 69–77. DOI
 Netzeva, T. I., Pavan, M., and Worth, A. P. (2008). Review of (Quantitative) Structure–Activity Relationships for Acute Aquatic Toxicity. QSAR Comb. Sci., 27(1), 7790. DOI
 Escher, B. I., and Hermens, J. L. M. (2002). Modes of Action in Ecotoxicology: Their Role in Body Burdens, Species Sensitivity, QSARs, and Mixture Effects. Environ. Sci. Technol., 36(20), 42014217. DOI
 Devillers, J. and Devillers, H. (2009). Prediction of acute mammalian toxicity from QSARs and interspecies correlations. SAR QSAR Environ. Res., 20(56), 467500. DOI
 Raevskii, O. A. and Grigor’ev, V. Yu. (1999). Lipophilicity of organic compounds: a quantitative description based on the concepts of polarizability and acceptor ability to hydrogen bond formation. Pharm. Chem. J., 33(5), 274277. DOI
 Markizova, N. F., Grebenyuk, A. N., Bašarin, V. A., & Bonitenko, E. J. (2004). Spirty: Toksikologia dlja vračej. VoennoMedicinskaja Akademie, SanktPeterburg.