QSAR model

In previous studies, we have found our hierarchical QSAR approach to be successful in predicting properties, activities, and toxicities including  

In this section, we provide a few examples of our FQQSAR studies involving heterocycHc compoimds, which are important both as drugs and toxicants. 

Gomplex field-based 3-D QSAR models have also been applied to the problem of predicting hERG activity. Gavalli ef al. [85] used a CoMFA model, as previously discussed. Pearlstein ef al. [89] modeled a set of sertindole analogs using compara- 

Multivariate models using neural networks, support vector machines and least median squares regression have been used to predict hERG activity [96-98]. These types of models function more as computational black box assays. 

The HlV-1 protease is responsible for processing the protein precursors to the enzymes (integrase, protease and reverse transcriptase) and the structural proteins of the HIV-1 virus. Maw and Hall found that topological indices provide rehable QSAR models for the IC50 data of 32 HIV-1 protease inhibitors [29]. The best QSAR model, with r = 0.86, s=0.60 and q = 0.79, was obtained with the shape index Ka, the connechvity index the sum of HE-state indices for ah groups that act as 

Derivatives of (S) N-[(l-ethyl-2-pyrrohdinyl)methyl]-6-methoxy benzamide 3 are dopamine D2 receptor antagonists. Samanta et al. obtained the following MLR QSAR for 49 derivatives with the general structure 3 [30]  

The neuropeptide Y (NPY) belongs to a family of peptides that includes peptide YY and pancreatic polypeptide, and it is associated with several diseases such as asthma, immune system disorders, inflammatory diseases, anxiety, depression and diabetes mellitus. NPY is found in the central and peripheral nervous system, and its biological functions are mediated by interactions with five receptor sub-types, i.e. Yl, Y2, Y4, Y5 and Y6. Several studies indicate that the feeding behavior is influenced by interactions between NPY and Yl and Y5. Deswal and Roy used Cerius descriptors and genetic function approximation QSAR to investigate the structural determinants for the inhibition potency of 24 compounds with the general structure 4 for the NPY Y5 receptor [31]. The best QSAR (H = 0.720, 

3loo = 0.616, F=12.2) was obtained with four indices, i.e. the E-state index for a N— group SsssN, the molecular connectivity index the area of the molecule projected on the XZ plane ShadowXZ, and the AlogP atom type count AtypeCS. 

The plCso values predicted for a test set of six compounds have a good correlation with the experimental values, r = 0.706, indicating that the QSAR model is stable and reliable. 

---

Hopfinger et al. [53, 54] have constructed 3D-QSAR models with the 4D-QSAR analysis formahsm. This formalism allows both conformational flexibility and freedom of alignment by ensemble averaging, i.e., the fourth dimension is the dimension of ensemble sampling. The 4D-QSAR analysis can be seen as the evolution of Molecular Shape Analysis [55, 56]. 

Activity prediction is based on a list of models (i.e., QSAR models, pharmacophore models, etc.) that are maintained on the server. There is a second level of access so that only authorized users may be allowed to add or delete model entries. 

The information in the structures and known activity data is good enough to create a QSAR model with a confidence of 75% ... 

The variable selection methods have been also adopted for region selection in the area of 3D QSAR. For example, GOLPE [31] was developed with chemometric principles and q2-GRS [32] was developed based on independent CoMFA analyses of small areas of near-molecular space to address the issue of optimal region selection in CoMFA analysis. Both of these methods have been shown to improve the QSAR models compared to original CoMFA technique. 

Shen M, Beguin C, Golbraikh A, Stables JP, Kohn H, Tropsha A. Application of predictive QSAR models to database mining identification and experimental validation of novel anticonvulsant compounds. J Med Chem 2004 47(9) 2356-64. 

Svetnik V, Liaw A, Tong C, Culberson JC, Sheridan RP, Feuston BP. Random forest a classification and regression tool for compound classification and QSAR modeling. J Chem Inf Comput Sci 2003 43(6) 1947-58. 

Kovatdieva A, Golbraikh A, Oloff S, Feng J, Zheng W, Tropsha A. QSAR modeling of datasets with enantioselective compounds using chirality sensitive molecular descriptors. SAR QSAR Environ Re. 2005 16(l-2) 93-102. 

MetaDrug Metabolism database. Metabolite prediction. Metabolite prioritization, QSAR models for enzymes, transporters and network building algorithms for Systems-ADME/Tox www.genego.com... 

QSAR modeling. Therefore considerably larger and more consistent data sets for each enzyme will be required in future to increase the predictive scope of such models. The evaluation of any rule-based metabolite software with a diverse array of molecules will indicate that it is possible to generate many more metabolites than have been identified in the literature for the respective molecules to date, which could also reflect the sensitivity of analytical methods at the time of publishing the data. In such cases, efficient machine learning algorithms will be necessary to indicate which of the metabolites are relevant and will be likely to be observed under the given experimental conditions. 

Livingstone [24] has given a number of recommendations for successful QSAR modeling ... 

MLR is the most widely used of the QSAR modeling techniques. Walker et al. [15] have published guidelines for the development and use of MLR-based QSARs, and Cronin and Schultz [41] have discussed their potential pitfalls. 

A QSAR for which the standard error of each descriptor is given concerns the bradycardic effect of a series of tetraalkylbispidines [47]. The QSAR models the selectivity between the desired bradycardic effect and the adverse contractile effect. It is important, in assessing and modeling drug toxicity, that the toxic effect is assessed relative to the desired effect as described above. The QSAR developed for the selectivity of the tetraalkylbispidines was ... 

A key requirement of QSAR is that the compounds used in the modeling and prediction processes should have the same mechanism of action, and for this reason most QSAR studies are made with congeneric series of compounds. However, if a diverse set of compounds can reasonably be assumed to have the same mechanism of action, QSAR modeling can justihably be carried out. For example, Dearden et al. [43] developed a QSAR for the ratio of brain levels of 22 very diverse drugs in the wild-type mouse and the P-glycoprotein knockout mouse (R+/ ) ... 

So far, we have considered the QSAR modeling of continuous biological data, that is, where the toxicity value is a number such as an LD50. However, some data are not continuous but are binary (e.g., toxic/ nontoxic) a common example is carcinogenicity, for which test results are almost invariably reported in this way. Clearly, one cannot perform, say, MLR on such classification data (although a method called fuzzy adaptive least squares [70] can be used). A number of methods are available for the modeling of classification data. 

Livingstone DJ. Building QSAR models a practical guide. In Cronin MTD, Livingstone DJ, editors. Predicting chemical toxicity and fate. Boca Raton CRC Press, 2004. p. 151-70. 

Dearden JC, Netzeva TI. QSAR modelling of hERG potassium channel inhibition with low-dimensional descriptors. I Pharm Pharmacol 2004 56 Suppl S-82. 

Devillers J. A general QSAR model for predicting the acute toxicity of pesticides to Lepomis macrochirus. SAR QSAR Environ Res 2001 11 397-417. 

The Danish EPA has developed an advisory list for self-classification of dangerous substances including 20 624 substances. The substances have been identified by means of QSAR models (Quantitative Structure-Activity Relationship) as having acute oral toxicity, sensitization, mutagenicity, carcinogenicity, and/or danger to the aquatic environment. 

---