Multiple Linear Regression Analysis and Optimal Descriptors: Predicting the Cholesteryl Ester Transfer Protein Inhibition Activity

Abstract

Quantitative structure -activity relationships have been developed for a set of 40 halogenated substituted N-benzyl-N-phenyl aminoalcohol compounds. IC 50 values for Cholesteryl Ester Transfer Protein (CETP) inhibition activity for these compounds expressed in log units have been modeled by Multiple Linear Regression Analysis (MLRA) based on descriptors generated by DRAGON software and optimal descriptors approach. Forty benzene derivatives have been divided into training (n ¼ 20) and test (n ¼ 20) sets. In the case of the MLRA, the three-variable model has the best predictive potential. Comparison of the quality of MLRA and optimal descriptor models shows that the predictive potential of the one-variable model, based on the optimal descriptor, provides a low correlation coefficient in the training set. However, this one-variable model performs better for the test set when compared to the two-and three-variable MLRA models.

Simplified Molecular Input Line Entry System‐Based Optimal Descriptors: Quantitative Structure–Activity Relationship Modeling Mutagenicity of Nitrated Polycyclic Aromatic Hydrocarbons

“…It is well known that the increase of the number of descriptors improves statistical quality of a model that is obtained with the multiple linear regression analysis (MLRA) for the training set (15). However, it can be accompanied by decrease of the statistical quality of this model for an external test set (16–18). For the toxicity towards rats (16,17) and the inhibition of the protein of cholesteryl ester transfer (18), the robust MLRA predictions are the three‐variable models or even two‐variable models.…”

“…However, it can be accompanied by decrease of the statistical quality of this model for an external test set (16–18). For the toxicity towards rats (16,17) and the inhibition of the protein of cholesteryl ester transfer (18), the robust MLRA predictions are the three‐variable models or even two‐variable models. It should be noted that in above‐mentioned studies, one‐variable models which are calculated with optimal descriptors (either based on molecular graph or based on SMILES) are better than the MLRA models.…”

QSPR modeling of enthalpies of formation for organometallic compounds by SMART‐based optimal descriptors

“…The numerical data on the CW( S k ) was calculated by the Monte Carlo method optimization procedure 18–23. The CW( S k ) provide the maximum of the correlation coefficient between the DCW and Δ f H 0 .…”

“…There is a tendency to use SMILES in databases on physicochemical and biochemical parameters 15, 16. The easy access to these data is a motivation to construct the QSPR/QSAR models calculated directly with the SMILES: there are examples of using of this approach for the QSPR/QSAR 17–23…”

Use of quantitative structure–enantioselective retention relationship for the liquid chromatography chiral separation prediction of the series of pyrrolidin‐2‐one compounds

“…Authors of this study previously developed various predictive models for different kinds of physico-chemical properties and biological activities, including degradation rates, 20 solubility of fullerene in various sol-vents (for better separation), 20,21,22 estrogenic activity, antiarrhythmic, and so on. [23][24][25][26][27] In this article, the QSERR study has been performed on a series of chiral pyrrolidin-2-ones to find the main factors affecting the enantioselectivity. For this purpose, a number of generated structural descriptors and quantum-chemical descriptors have been applied and the ones that are most responsible for the enantioselectivity were defined.…”