Detection and classification of organophosphate nerve agent simulants using support vector machines with multiarray sensors

The need for rapid and accurate detection systems is expanding and the utilization of cross-reactive sensor arrays to detect chemical warfare agents in conjunction with novel computational techniques may prove to be a potential solution to this challenge. We have investigated the detection, prediction, and classification of various organophosphate (OP) nerve agent simulants using sensor arrays with a novel learning scheme known as support vector machines (SVMs). The OPs tested include parathion, malathion, dichlorvos, trichlorfon, paraoxon, and diazinon. A new data reduction software program was written in MATLAB V. 6.1 to extract steady-state and kinetic data from the sensor arrays. The program also creates training sets by mixing and randomly sorting any combination of data categories into both positive and negative cases. The resulting signals were fed into SVM software for "pairwise" and "one" vs all classification. Experimental results for this new paradigm show a significant increase in classification accuracy when compared to artificial neural networks (ANNs). Three kernels, the S2000, the polynomial, and the Gaussian radial basis function (RBF), were tested and compared to the ANN. The following measures of performance were considered in the pairwise classification: receiver operating curve (ROC) Az indices, specificities, and positive predictive values (PPVs). The ROC Az) values, specifities, and PPVs increases ranged from 5% to 25%, 108% to 204%, and 13% to 54%, respectively, in all OP pairs studied when compared to the ANN baseline. Dichlorvos, trichlorfon, and paraoxon were perfectly predicted. Positive prediction for malathion was 95%.     

Jointly handling potency and toxicity of antimicrobial peptidomimetics by simple rules from desirability theory and chemoinformatics

Today, emerging and increasing resistance to antibiotics has become a threat to public health worldwide. Antimicrobial peptides have unique action mechanisms making them an attractive therapeutic prospect to be applied against resistant bacteria. However, the major drawback is related with their high hemolytic activity which cancels out the safety requirements for a human antibiotic. Therefore, additional efforts are needed to develop new antimicrobial peptides that possess a greater potency for bacterial cells and less or no toxicity over erythrocytes. In this paper, we introduce a practical approach to simultaneously deal with these two conflicting properties. The convergence of machine learning techniques and desirability theory allowed us to derive a simple, predictive, and interpretable multicriteria classification rule for simultaneously handling the antibacterial and hemolytic properties of a set of cyclic β-hairpin cationic peptidomimetics (Cβ-HCPs). The multicriteria classification rule exhibited a prediction accuracy of about 80% on training and external validation sets. Results from an additional concordance test have shown an excellent agreement between the multicriteria classification rule predictions and the predictions from independent classifiers for complementary antibacterial and hemolytic activities, respectively, evidencing the reliability of the multicriteria classification rule. The rule was also consistent with the general mode of action of cationic peptides pointing out its biophysical relevance. We also propose a multicriteria virtual screening strategy based on the joint use of the multicriteria classification rule, desirability, similarity, and chemometrics concepts. The ability of such a virtual screening strategy to prioritize selective (nonhemolytic) antibacterial Cβ-HCPs was assessed and challenged for their predictivity regarding the training, validation, and overall data. In doing so, we were able to rank a selective antibacterial Cβ-HCP earlier than a biologically inactive or nonselective antibacterial Cβ-HCP with a probability of ca. 0.9. Our results thus indicate that promising chemoinformatics tools were obtained by considering both the multicriteria classification rule and the virtual screening strategy, which could, for instance, be used to aid the discovery and development of potent and nontoxic antimicrobial peptides.