Optimization Of Correlation Weights Research Articles

QSAR study of dihydrofolate reductase inhibitors activities based on optimization of correlation weights of local graph invariants

The aim of this work is modeling the activities of 67 E. coli dihydrofolate reductase (DHFR) inhibitors 2,4-diamino-5-(substituted benzyl) pyramidines by optimization of correlation weights of local graph invariants (cwlgis), assisting genetic algorithm (GA). The relationship between the descriptors and molecular activities was modeled using radial basis function partial least squares (rbfPLS). Proper relation between activities and the optimized molecular descriptors were shown, R training = 0.92 and r test = 0.85. The results indicate promising potential for the optimization of a correlation weights scheme. It is the first application of GA in optimization of cwlgis. The study also shows the ability of rbfPLS, combined with GA, in a factor-based nonlinear QSAR modeling using simple flexible descriptors.

QSPR-modeling of oligophenylene melting points

QSPR-modeling of melting temperatures has been performed for oligophenylenes of diverse structures and different molecular weights. It has been shown that optimization of correlation weights of graph local invariants can be used for prediction of melting points of oligomers. The best statistical quality of the model is observed for correlation weights calculated using third-order Morgan degrees.

QSPR Modeling of Gibbs Free Energy of Organic Compounds by Weighting of Nearest Neighboring Codes

We examine the encoding of chemical structure of organic compounds by Labeled Hydrogen-Filled Graphs (LHFGs). Quantitative Structure-Property Relationships (QSPR) for a representative set of 150 organic molecules have been derived by means of the optimization of correlation weights of local invariants of the LHFGs. We have tested as local invariants Morgan extended connectivity of zero- and first order, numbers of path of length 2 (P2) and valence shells of distance of 2 (S2) associated with each atom in the molecular structure, and the Nearest Neighboring Codes (NNC). The best statistical characteristics for the Gibbs free energy has been obtained for the NNC weighting. Statistical parameters corresponding to this model are the following n = 100, r2 = 0.9974, s = 5.136 kJ/mol, F = 38319 (training set); n = 50, r2 = 0.9990, s = 3.405 kJ/mol, F = 48717 (test set). Some possible further developments are pointed out.

QSPR modeling of the water solubility of diverse functional aliphatic compounds by optimization of correlation weights of local graph invariants

The optimization of correlation weights scheme was used to model the water solubility (ln S) of diverse functional aliphatic compounds (n = 193). The optimized descriptor formulated based on the data of a training set (n = 96) generated statistically acceptable relations for the training set (r2 = 0.987), test set (n = 97; r2 = 0.986) and combined set (r2 = 0.987). When the relation of ln S values with the optimized molecular descriptor formulated based on the data of the training set was used for the calculation of ln S values of the training set, r(pred)2 value was found to be satisfactory (0.988), which is indicative of the predictive potential of the scheme. The results indicate the promising potential of the optimization of correlation weights scheme in modeling studies.

QSAR Modeling ANTI-HIV-1 Activities by Optimization of Correlation Weights of Local Graph Invariants

Results of using descriptors calculated with the correlation weights (CWs) of local graph invariants for modeling of anti-HIV-1 potencies of two groups of reverse transcriptase (RT) inhibitors are reported. Presence of different chemical elements in molecular structure of the inhibitors and the presence of Morgan extended connectivity values of zeroth-, first- and second order have been examined as local graph invariants in the labeled hydrogen-filled graphs. By Monte Carlo method optimization procedure, values of the CWs which produce as large values as possible of correlation coefficient between the numerical data on the anti-HIV-1 potencies and values of the descriptors on the training set have been computed. The model of the anti-HIV-1 activity obtained with compounds of training set by means of optimization of correlation weights of presence of chemical elements together with the presence of Morgan extended connectivity of first order is reasonable well model for the prediction of endpoints under consideration for compounds of the test set.

Structure–Toxicity Relationships for Aliphatic Compounds Based on Correlation Weighting of Local Graph Invariants

Quantitative Structure-Activity Relationships based on molecular descriptors calculated with Correlation Weights of Local Graph Invariants were developed to model the toxicity of aliphatic compounds to the 50% population growth inhibition. The relationships were computed on the basis of Labeled Hydrogen- Filled Graphs and correlation weights were obtained by an optimization to render as large as possible correlation coefficients between log(IGC50-1) and descriptors calculated with correlation weights. Morgan extended connectivity indices of zero, first, and second orders, paths of lengths two and three and valence shells of second and third ranges have been tested as local invariants of the Labeled Hydrogen-Filled Graphs. The best quantitative relationship obtained from the optimization of correlation weights is that one based on the valence shell of range two. First, second, and third order fitting equations were determined and statistical results are better than other similar data for the same molecular set.

QSPR Modeling of Stability of Complexes of Adenosine Phosphate Derivatives with Metals Absent from the Complexes of the Teaching Access

A descriptor calculated from correlation weights of the main quantum number, orbital quantum number, number of electrons on the atomic orbital, and Morgan degrees of the second-order vertices in the graph of atomic orbitals (GAO) was proposed. Using the optimization of correlation weights (OCW) of the listed local invariants of GAO, their values were found at which the correlation coefficient between the descriptor and stability constants is maximum for teaching accesses. The quality of models of stability constants of the complexes were verified by reference accesses. The results of the series of OCW experiments show that the proposed approach allows one to predict the stability constants of the Ca2+, Cu2+, and Zn2+complexes with adenosine phosphate derivatives using teaching accesses containing no complexes of these metals.