Rafal Gieleciak

Modeling robust QSAR

Quantitative Structure Activity Relationship (QSAR) is a term describing a variety of approaches that are of substantial interest for chemistry. This method can be defined as indirect molecular design by the iterative sampling of the chemical compounds space to optimize a certain property and thus indirectly design the molecular structure having this property. However, modeling the interactions of chemical molecules in biological systems provides highly noisy data, which make predictions a roulette risk. In this paper we briefly review the origins for this noise, particularly in multidimensional QSAR. This was classified as the data, superimposition, molecular similarity, conformational, and molecular recognition noise. We also indicated possible robust answers that can improve modeling and predictive ability of QSAR, especially the self-organizing mapping of molecular objects, in particular, the molecular surfaces, a method that was brought into chemistry by Gasteiger and Zupan.

The Comparative Molecular Surface Analysis (CoMSA) with Modified Uniformative Variable Elimination-PLS (UVE-PLS) Method: Application to the Steroids Binding the Aromatase Enzyme

The application of the CoMSA method to analyze 3D QSAR of 50 steroid aromatase inhibitors is described. The 3D QSAR model obtained, reaching a value of cross-validated q(2) = 0.96 (s = 0.31), significantly outperforms those reported in the literature for the CoMFA or CoSA (CoSASA). It is shown that the Uniformative Variable Elimination UVE-PLS or modified iterative UVE procedure (IVE-PLS) can be used for indicating the regions contributing to the binding activity. Thus, after separating the series into two groups of the training and test molecules quite correct external predictions result from the processing of the training set. We proved that the procedure of the data elimination provides stable results, if tested in 50 random runs of the IVE-PLS-CoMSA with different training/test sets. Depending upon the procedure used the quality of the predictions for 25 test molecules is given by SDEP = sum(y(pred)-y(obs))(2)/n)(1/2) = 0.321 - 0.782.

The comparative molecular surface analysis (COMSA): A nongrid 3D QSAR method by a coupled neural network and PLS system: Predicting pKa values of benzoic and alkanoic acids

A self-organizing neural network was used to design a novel method capable of the quantitative prediction of molecular properties. The method is based on the comparison of molecular surfaces performed by the coupled neural network and PLS system. Unlike CoMFA and related methods it does not compare the properties describing a discrete set of points but the average property values calculated for a certain area of the molecular surface. It has been found that the results of the PLS analysis of the series of the comparative matrices of the molecular electrostatic potential (MEP) are quite stable. Also the results only slightly depend on such parameters as the number of points sampled at the molecular surface (D) or a winning distance (MD) of the self-organizing neurons. The influence of these parameters for modeling the effects limited by steric and electronic effects was determined and the pK(a) values of the ortho-, meta-, and para- (o-, m-, p-) analogues of benzoic acid and selected alkanoic acids were predicted. We generally found that for the series analyzed CoMSA gave better models than CoMFA.

Probability Issues in Molecular Design: Predictive and Modeling Ability in 3D-QSAR Schemes

In the current work we investigated 3D-QSAR data by the use of the coupled leave-several-out (LSO) and leave-one-out (LOO) cross-validation (CV) procedures. We verified the above mentioned scheme using both simulated data and real 3D QSAR data describing a series of CoMFA steroids, heterocyclic azo dyes and styrylquinoline HIV integrase inhibitors. Unlike in standard analyses, this technique characterizes individual method not by a single performance metrics but screens a whole possible modeling space by sampling different molecules into the training and test sets, respectively. This allowed us for the discussion of the information included in the estimators validating cross-validation procedures, as well as the comparison of the efficiency of several 3D QSAR schemes, in particular, Comparative Molecular Field Analysis (CoMFA) and Comparative Molecular Surface Analysis (CoMSA). Moreover, it allows one to acquire some general knowledge about predictive and modeling ability in 3D QSAR method.

Comparative molecular surface analysis (CoMSA) for modeling dye-fiber affinities of the azo and anthraquinone dyes.

Despite recent investigations aimed at modeling 3D QSAR for dye molecules a controversy still exists: can a pharamacophore hypothesis be used for such purposes. In the present publication we reported on the application of the CoMSA method for modeling 3D QSAR of azo and anthraquinone dyes. We obtained very predictive models, which significantly outperform those reported in the previous CoMFA studies, especially for the azo dyes. Our results proved the previous conclusion that steric requirements are far less pronounced for the cellulose cavities than for the classical drug receptor. Moreover, our results indicate that all molecular surface segments are important for dye-fiber interactions, which also makes an important difference in relation to the classical drug pharmacophore. On the other hand, high predictivity of the CoMSA models indicates that a pharmacophore concept is suitable for the description of the dye-fiber interactions. However, this pharmacophore must substantially differ from the drug pharmacophore used for the illustration of the drug-receptor interactions. From a theoretical point of view dye-cellulose interactions can be an interesting case in which shape decides the activity rules not by the steric repulsion but as a cofactor determining the electrostatic potential distribution.

Self-organizing Neural Networks for Modeling Robust 3D and 4D QSAR: Application to Dihydrofolate Reductase Inhibitors

We have used SOM and grid 3D and 4D QSAR schemes for modeling the activity of a series of dihydrofolate reductase inhibitors. Careful analysis of the performance and external predictivities proves that this method can provide an efficient inhibition model.

Comparative molecular surface analysis (CoMSA) for virtual combinatorial library screening of styrylquinoline HIV-1 blocking agents.

We used comparative molecular surface analysis to design molecules for the synthesis as part of the search for new HIV-1 integrase inhibitors. We analyzed the virtual combinatorial library (VCL) constituted from various moieties of styrylquinoline and styrylquinazoline inhibitors. Since imines can be applied in a strategy of dynamic combinatorial chemistry (DCC), we also tested similar compounds in which the -C=N- or -N=C- linker connected the heteroaromatic and aromatic moieties. We then used principal component analysis (PCA) or self-organizing maps (SOM), namely, the Kohonen neural networks to obtain a clustering plot analyzing the diversity of the VCL formed. Previously synthesized compounds of known activity, used as molecular probes, were projected onto this plot, which provided a set of promising virtual drugs. Moreover, we further modified the above mentioned VCL to include the single bond linker -C-N- or -N-C-. This allowed increasing compound stability but expanded also the diversity between the available molecular probes and virtual targets. The application of the CoMSA with SOM indicated important differences between such compounds and active molecular probes. We synthesized such compounds to verify the computational predictions.

Comparative molecular surface analysis: A novel tool for drug design and molecular diversity studies

The application of the SOM network in drug design and molecular diversity is discussed. In particular, examples of the applications of the Comparative Molecular Surface Analysis (CoMSA) are reviewed. Molecular surface is a fuzzy category, inspired by the macroscopic world, which has no unique equivalent in the molecular scale. However, it is somewhere near the area where the molecular recognition processes are taking place. Consequently, the methods that analyze this region promise better efficiency than procedures that are based on uniform grids. An important advantage of the CoMSA method is the possibility for the generation of fuzzy molecular representations together with its ability to discover such aspects of molecular similarity that can be easily overlooked by a chemist. The ability for data compression is a further advantage. It has also been shown that the fast processing of the comparative Kohonen mapping enables one to implement this method in the field of molecular diversity.