Kunal Roy

On Two Novel Parameters for Validation of Predictive QSAR Models

Validation is a crucial aspect of quantitative structure–activity relationship (QSAR) modeling. The present paper shows that traditionally used validation parameters (leave-one-out Q2 for internal validation and predictive R2 for external validation) may be supplemented with two novel parameters rm2 and Rp2 for a stricter test of validation. The parameter rm2(overall) penalizes a model for large differences between observed and predicted values of the compounds of the whole set (considering both training and test sets) while the parameter Rp2 penalizes model R2 for large differences between determination coefficient of nonrandom model and square of mean correlation coefficient of random models in case of a randomization test. Two other variants of rm2 parameter, rm2(LOO) and rm2(test), penalize a model more strictly than Q2 and R2pred respectively. Three different data sets of moderate to large size have been used to develop multiple models in order to indicate the suitability of the novel parameters in QSAR studies. The results show that in many cases the developed models could satisfy the requirements of conventional parameters (Q2 and R2pred) but fail to achieve the required values for the novel parameters rm2 and Rp2. Moreover, these parameters also help in identifying the best models from among a set of comparable models. Thus, a test for these two parameters is suggested to be a more stringent requirement than the traditional validation parameters to decide acceptability of a predictive QSAR model, especially when a regulatory decision is involved.

Some case studies on application of “rm2” metrics for judging quality of quantitative structure–activity relationship predictions: Emphasis on scaling of response data

Quantitative structure–activity relationship (QSAR) techniques have found wide application in the fields of drug design, property modeling, and toxicity prediction of untested chemicals. A rigorous validation of the developed models plays the key role for their successful application in prediction for new compounds. The rm2 metrics introduced by Roy et al. have been extensively used by different research groups for validation of regression‐based QSAR models. This concept has been further advanced here with introduction of scaling of response data prior to computation of rm2. Further, a web application (accessible from http://aptsoftware.co.in/rmsquare/ and http://203.200.173.43:8080/rmsquare/) for calculation of the rm2 metrics has been introduced here. The present study reports that the web application can be easily used for computation of rm2 metrics provided observed and QSAR‐predicted data for a set of compounds are available. Further, scaling of response data is recommended prior to rm2 calculation.

Exploring quantitative structure–activity relationship studies of antioxidant phenolic compounds obtained from traditional Chinese medicinal plants

In the present work, quantitative structure–activity relationship (QSAR) models have been built for a wide variety of antioxidant phenolic compounds obtained from traditional Chinese medicinal plants, with their Trolox equivalent antioxidant capacity measured using 1,1-diphenyl-2-picrylhydrazyl (DPPH) radical and 2,2′-azinobis-(3-ethylbenzothiazoline-6-sulphonic acid) radical (ABTS√+) assay methods. Non-linear models obtained using genetic partial least-squares technique were acceptable both in terms of internal and external predictivity. Validation of developed models using metrics and randomisation technique yielded results indicating the predictivity and robustness, respectively, of the developed models. The models signify that the presence of ketonic oxygen within the molecular structure favours their antioxidant activity. In addition, the number of hydroxyl groups, extent of branching, degree of methoxylation and the number of methyl and methylene substituents also dictate the antioxidant activity of these molecules. Thus, the QSAR models developed here can be utilised for the antioxidant activity prediction of a new series of molecules.

On some aspects of validation of predictive quantitative structure–activity relationship models

The success of any quantitative structure–activity relationship model depends on the accuracy of the input data, selection of appropriate descriptors and statistical tools and, most importantly, the validation of the developed model. Validation is the process by which the reliability and relevance of a procedure are established for a specific purpose. This review focuses on the importance of validation of quantitative structure–activity relationship models and different methods of validation. Some important issues, such as internal versus external validation, method of selection of training set compounds and training set size, applicability domain, variable selection and suitable parameters to indicate external predictivity, are also discussed.

On various metrics used for validation of predictive QSAR models with applications in virtual screening and focused library design.

Quantitative structure-activity relationships (QSARs) have important applications in drug discovery research, environmental fate modeling, property prediction, etc. Validation has been recognized as a very important step for QSAR model development. As one of the important objectives of QSAR modeling is to predict activity/property/toxicity of new chemicals falling within the domain of applicability of the developed models and QSARs are being used for regulatory decisions, checking reliability of the models and confidence of their predictions is a very important aspect, which can be judged during the validation process. One prime application of a statistically significant QSAR model is virtual screening for molecules with improved potency based on the pharmacophoric features and the descriptors appearing in the QSAR model. Validated QSAR models may also be utilized for design of focused libraries which may be subsequently screened for the selection of hits. The present review focuses on various metrics used for validation of predictive QSAR models together with an overview of the application of QSAR models in the fields of virtual screening and focused library design for diverse series of compounds with citation of some recent examples.

Advances in QSPR/QSTR models of ionic liquids for the design of greener solvents of the future

In order to protect the life of all creatures living in the environment, the toxicity arising from various hazardous chemicals must be controlled. This imposes a serious responsibility on different chemical, pharmaceutical, and other biological industries to produce less harmful chemicals. Among various international initiatives on harmful aspects of chemicals, the ‘Green Chemistry’ ideology appears to be one of the most highlighted concepts that focus on the use of eco-friendly chemicals. Ionic liquids are a comparatively new addition to the huge garrison of chemical compounds released from the industry. Extensive research on ionic liquids in the past decade has shown them to be highly useful chemicals with a good degree of thermal and chemical stability, appreciable task specificity and minimal environmental release resulting in a notion of ‘green chemical’. However, studies have also shown that ionic liquids are not intrinsically non-toxic agents and can pose severe degree of toxicity as well as the risk of bioaccumulation depending upon their structural components. Moreover, ionic liquids possess issues of waste generation during synthesis as well as separation problems. Predictive quantitative structure–activity relationship (QSAR) models constitute a rational opportunity to explore the structural attributes of ionic liquids towards various physicochemical and toxicological endpoints and thereby leading to the design of environmentally more benevolent analogues with higher process selectivity. Such studies on ionic liquids have been less extensive compared to other industrial chemicals. The present review attempts to summarize different QSAR studies performed on these chemicals and also highlights the safety, health and environmental issues along with the application specificity on the dogma of ‘green chemistry’.

Comparative chemometric modeling of cytochrome 3A4 inhibitory activity of structurally diverse compounds using stepwise MLR, FA-MLR, PLS, GFA, G/PLS and ANN techniques

Twenty-eight structurally diverse cytochrome 3A4 (CYP3A4) inhibitors have been subjected to quantitative structure-activity relationship (QSAR) studies. The analyses were performed with electronic, spatial, topological, and thermodynamic descriptors calculated using Cerius 2 version 10 software. The statistical tools used were linear [multiple linear regression with factor analysis as preprocessing step (FA-MLR), stepwise MLR, partial least squares (PLS), genetic function algorithm (GFA), genetic PLS (G/PLS)] and non-linear methods [artificial neural network (ANN)]. All the five linear modeling methods indicate the importance of n-octanol/water partition coefficient (logP) along with different topological and electronic parameters. The best model obtained from the training set (stepwise regression) based on highest external predictive R(2) value and lowest RMSEP value also showed good internal predictive power. Other models like FA-MLR, PLS, GFA and G/PLS are also of statistically significant internal and external validation characteristics. The best model [according to r(m)(2) for the test set, as defined by P.P. Roy, K. Roy, QSAR Comb. Sci. 27 (2008) 302-313] obtained from ANN showed a good r(2) value (determination coefficient between observed and predicted values) for the test set compounds, which was superior to those of other statistical models except the stepwise regression derived model. However, based upon the r(m)(2) value (test set), which penalizes a model for large differences between observed and predicted values, the stepwise MLR model was found to be inferior to other methods except PLS. Considering r(m)(2) value for the whole set, the G/PLS derived model appears to be the best predictive model for this data set. For choosing the best predictive model from among comparable models, r(m)(2) for the whole set calculated based on leave-one-out predicted values of the training set and model-derived predicted values for the test set compounds is suggested to be a good criterion.

QSAR modeling of toxicity of diverse organic chemicals to Daphnia magna using 2D and 3D descriptors.

One of the major economic alternatives to experimental toxicity testing is the use of quantitative structure-activity relationships (QSARs) which are used in formulating regulatory decisions of environmental protection agencies. In this background, we have modeled a large diverse group of 297 chemicals for their toxicity to Daphnia magna using mechanistically interpretable descriptors. Three-dimensional (3D) (electronic and spatial) and two-dimensional (2D) (topological and information content indices) descriptors along with physicochemical parameter logK(o/w) (n-octanol/water partition coefficient) and structural descriptors were used as predictor variables. The QSAR models were developed by stepwise multiple linear regression (MLR), partial least squares (PLS), genetic function approximation (GFA), and genetic PLS (G/PLS). All the models were validated internally and externally. Among several models developed using different chemometric tools, the best model based on both internal and external validation characteristics was a PLS equation with 7 descriptors and three latent variables explaining 67.8% leave-one-out predicted variance and 74.1% external predicted variance. The PLS model suggests that higher lipophilicity and electrophilicity, less negative charge surface area and presence of ether linkage, hydrogen bond donor groups and acetylenic carbons are responsible for greater toxicity of chemicals. The developed model may be used for prediction of toxicity, safety and risk assessment of chemicals to achieve better ecotoxicological management and prevent adverse health consequences.

QSTR with extended topochemical atom indices. 2. Fish toxicity of substituted benzenes.

Considering the importance of quantitative structure-toxicity relationship (QSTR) studies in the field of aquatic toxicology from the viewpoint of ecological safety assessment, fish toxicity of various benzene derivatives has been modeled by the multiple regression technique using recently introduced extended topochemical atom (ETA) indices. The toxicity data have also been modeled using other selected topological descriptors and physicochemical variables, and the best ETA model has been compared to the non-ETA ones. Principal component factor analysis was used as the data preprocessing step to reduce the dimensionality of the data matrix and identify the important variables that are devoid of collinearities. All-possible-subsets regression was also applied on the parameters to cross-check the variable selection for the best model. Multiple linear regression analyses show that the best non-ETA model involves 1chi, ALogP98, and LUMO (energy) as predictor variables and the quality of the relation is as follows: n = 92, Q2 = 0.718, Ra2 = 0.730, R2 = 0.738, R = 0.859, F = 82.8 (df 3, 88), s = 0.340. On the other hand, the best ETA model has the following quality: n = 92, Q2 = 0.865, Ra2 = 0.876, R2 = 0.885, R = 0.941, F = 92.6 (df 7, 84), s = 0.230. The ETA relations showed positive contributions of molecular bulk (size), chloro and hydroxy substitutions in the benzene ring, and the simultaneous presence of methyl and nitro substitutions to the toxicity. Further, the presence of fluoro and ether functionality, amino or nitro functionality in an otherwise unsubstituted ring, and nitro functionality that is ortho to a chloro substituent decreases toxicity. An attempt to use non-ETA descriptors along with ETA ones did not improve the quality in comparison to the best ETA model. Interestingly, the ETA model developed presently for the fish toxicity is better than the previously reported models on the same data set. Thus, it appears that ETA descriptors have significant potential in QSAR/QSPR/QSTR studies, which warrants extensive evaluation.

Nano-quantitative structure–activity relationship modeling using easily computable and interpretable descriptors for uptake of magnetofluorescent engineered nanoparticles in pancreatic cancer cells

As experimental evaluation of the safety of nanoparticles (NPs) is expensive and time-consuming, computational approaches have been found to be an efficient alternative for predicting the potential toxicity of new NPs before mass production. In this background, we have developed here a regression-based nano quantitative structure-activity relationship (nano-QSAR) model to establish statistically significant relationships between the measured cellular uptakes of 109 magnetofluorescent NPs in pancreatic cancer cells with their physical, chemical, and structural properties encoded within easily computable, interpretable and reproducible descriptors. The developed model was rigorously validated internally as well as externally with the application of the principles of Organization for Economic Cooperation and Development (OECD). The test for domain of applicability was also carried out for checking reliability of the predictions. Important fragments contributing to higher/lower cellular uptake of NPs were identified through critical analysis and interpretation of the developed model. Considering all these identified structural attributes, one can choose or design safe, economical and suitable surface modifiers for NPs. The presented approach provides rich information in the context of virtual screening of relevant NP libraries.