Determining the validity of a QSAR model--a classification approach

The determination of the validity of a QSAR model when applied to new compounds is an important concern in the field of QSAR and QSPR modeling. Various scoring techniques can be applied to specific types of models. We present a technique with which we can state whether a new compound will be well predicted by a previously built QSAR model. In this study we focus on linear regression models only, though the technique is general and could also be applied to other types of quantitative models. Our technique is based on a classification method that divides regression residuals from a previously generated model into a good class and bad class and then builds a classifier based on this division. The trained classifier is then used to determine the class of the residual for a new compound. We investigated the performance of a variety of classifiers, both linear and nonlinear. The technique was tested on two data sets from the literature and a hand built data set. The data sets selected covered both physical and biological properties and also presented the methodology with quantitative regression models of varying quality. The results indicate that this technique can determine whether a new compound will be well or poorly predicted with weighted success rates ranging from 73% to 94% for the best classifier.     

Fabrication of hollow silica microspheres utilizing a hydrothermal approach

Hollow silica microspheres(HSMSs) have been successfully fabricated via a facile hydrothermal route using D-glucose as the sacrificial template and sodium silicate powder as the silica precursor.The resulting silica hollow particles were characterized by scanning electron microscopy(SEM),transmission electron microscopy(TEM),X-ray diffraction(XRD),and infrared spectroscopy(IR).The surface area was determined using the BET method.SEM and TEM images exhibited micro-sized silica hollow particles with a size of ~1.5μm.     

Fabrication of hierarchical porous iron oxide films utilizing the Kirkendall effect

Hierarchical porous iron oxide films with different morphologies have for the first time been fabricated through hydrothermal reactions between an iron substrate and iodine powder in water or ethanol, which can be explained by a mechanism analogous to the Kirkendall effect.     

In-silico screening of high production volume chemicals for mutagenicity using the MCASE QSAR expert system

Computational screening is suggested as a way to set priorities for further testing of high production volume (HPV) chemicals for mutagenicity and other toxic endpoints. Results are presented for batch screening of 2484 HPV chemicals to predict their mutagenicity in Salmonella typhimurium (Ames test). The chemicals were tested against 15 databases for Salmonella strains TA100, TA1535, TA1537, TA97 and TA98, both with metabolic activation (using rat liver and hamster liver S9 mix test) and without metabolic activation. Of the 2484 chemicals, 1868 are predicted to be completely nonmutagenic in all of the 15 data modules and 39 chemicals were found to contain structural fragments outside the knowledge of the expert system and therefore suggested for further evaluation. The remaining 616 chemicals were found to contain different biophores (structural alerts) believed to be linked to mutagenicity. The chemicals were ranked indescending order according to their predicted mutagenic potential and the first 100 chemicals with highest mutagenicity scores are presented. The screening result offers hope that rapid and inexpensive computational methods can aid in prioritizing the testing of HPV chemicals, save time and animals and help to avoid needless expense.     

An ensemble approach for in silico prediction of Ames mutagenicity

In this paper, we evaluate three learning algorithms based on supervised projections for molecular activity prediction. Using an approach based on supervised projections of the input space to construct ensembles of classifiers, three algorithms were tested. We constructed the projections by considering only instances that were misclassified by a previous classifier using the hidden layer of an Artificial Neural Network. We applied a supervised linear projection of the input space using a Nonparametric Discriminant Analysis method. Finally, we projected onto a subspace that minimizes the weighted error for each step. Using these three methods to construct ensembles of classifiers for the in silico prediction of Ames mutagenicity, we demonstrated the improved behavior of our proposal compared to classical methods.     

Synthesis of tetra-ortho-substituted, phosphorus-containing and carbonyl-containing biaryls utilizing a Diels-Alder approach

The application of the Diels-Alder approach to biaryls (DAB) is described for the synthesis of tetra-ortho-substituted biaryl compounds containing orthogonally functionalized substituents. The syntheses of phosphorus-containing, disubstituted alkynes and carbonyl-containing, disubstituted alkynes were accomplished in two to three steps from commercially available reagents. Subsequent Diels-Alder cycloadditions with a range of oxygenated dienes yielded the target biaryls. Further functionalization through palladium-couplings is demonstrated on the phosphorus-containing biaryls. In addition, selective manipulation of each of the remaining ortho substituents on the phosphorus-containing biaryls is demonstrated. One of these phosphorus-containing derivatives is utilized as a highly active catalyst for Suzuki coupling. For the carbonyl-containing series, a wide range of dienophile substituents were screened including esters, ketones, and amides. The key Diels-Alder cycloadditions proceeded smoothly with the commercially available 1-methoxy-1,3-cyclohexadiene to yield the resultant tetra-ortho-substituted biaryls with excellent regioselectivity. The scope of the cycloaddition process was also explored on the carbonyl-containing dienophiles with a series of cyclic dienes. Acyclic dienes were also screened; however, they did not prove effective in the Diels-Alder process with the carbonyl-containing acetylenes. The ability to isolate enantiomerically pure biaryl atropisomers using a benzyl oxazolidinone is disclosed. Finally, the subsequent conversion to an axially chiral anilino alcohol is also reported.     

An evaluation of experimental design in QSAR modelling utilizing the k‐medoid clustering

A reliable selection of a representative subset of chemical compounds has been reported to be crucial for numerous tasks in computational chemistry and chemoinformatics. We investigated the usability of an approach on the basis of the k‐medoid algorithm for this task and in particular for experimental design and the split between training and validation set. We therefore compared the performance of models derived from such a selection to that of models derived using several other approaches, such as space‐filling design and D‐optimal design. We validated the performance on four datasets with different endpoints, representing toxicity, physicochemical properties and others. Compared with the models derived from the compounds selected by the other examined approaches, those derived with the k‐medoid selection show a high reliability for experimental design, as their performance was constantly among the best for all examined datasets. Of all the models derived with all examined approaches, those derived with the k‐medoid approach were the only ones that showed a significantly improved performance compared with a random selection, for all datasets, the whole examined range of selected compounds and for each dimensionality of the search space. Copyright © 2012 John Wiley & Sons, Ltd.     

A steroids QSAR approach based on approximate similarity measurements

A new QSAR method based on approximate similarity measurements is described in this paper. Approximate similarity is calculated using both the classical similarity based on the graph isomorphism and a distance computation between nonisomorphic subgraphs. The latter is carried out through a parametric function where different topological invariants can be considered. After optimizing the contribution of nonisomorphic distance to the new graph similarity, predictive models built with approximate similarity matrixes show higher predictive ability than those using traditional similarity matrixes. The new method has been applied to the prediction of steroids binding to the corticosteroid globulin receptor. The proposed model allows us to obtain valuable external predictions (r=0.82 and SEP=0.30) after training the model by cross-validation (Q2=0.84 and SECV=0.47). Slope and bias parameters are also given.     

New approach by Kriging models to problems in QSAR

Most models in quantitative structure and activity relationship (QSAR) research, proposed by various techniques such as ordinary least squares regression, principal components regression, partial least squares regression, and multivariate adaptive regression splines, involve a linear parametric part and a random error part. The random errors in those models are assumed to be independently identical distributed. However, the independence assumption is not reasonable in many cases. Some dependence among errors should be considered just like Kriging. It has been successfully used in computer experiments for modeling. The aim of this paper is to apply Kriging models to QSAR. Our experiments show that the Kriging models can significantly improve the performances of the models obtained by many existing methods.     

Prediction of human cytochrome P450 2B6‐substrate interactions using hierarchical support vector regression approach

The human cytochrome P450 2B6 can metabolize a number of clinical drugs. Inhibition of CYP2B6 by coadministered multiple drugs may lead to drug–drug interactions and undesired drug toxicity. The aim of this investigation is to develop an in silico model to predict the interactions between P450 2B6 and novel inhibitors using a novel hierarchical support vector regression (HSVR) approach, which simultaneously takes into account the coverage of applicability domain (AD) and the level of predictivity. Thirty‐seven molecules were deliberately selected and rigorously scrutinized from the literature data, of which 26 and 11 molecules were treated as the training set and the test set to generate the models and to validate the generated models, respectively. The generated HSVR model gave rise to an r² value of 0.97 for observed versus predicted pK_m values for the training set, a q² value of 0.93 by the 10‐fold cross‐validation, and an r² value of 0.82 for the test set. Additionally, the predicted results show that the HSVR model outperformed the individual local models, the global model, and the consensus model. Thus, this HSVR model provides an accurate tool for the prediction of human cytochrome P450 2B6‐substrate interactions and can be utilized as a primary filter to eliminate the potential selective inhibitor of CYP2B6. © 2008 Wiley Periodicals, Inc. J Comput Chem, 2009