Dragon descriptors

The %HIA, on a scale between 0 and 100%, for the same dataset was modeled by Deconinck et al. with multivariate adaptive regression splines (MARS) and a derived method two-step MARS (TMARS) [38]. Among other Dragon descriptors, the TMARS model included the Tig E-state topological parameter [25], and MARS included the maximal E-state negative variation. The average prediction error, which is 15.4% for MARS and 20.03% for TMARS, shows that the MARS model is more robust in modeling %H1A. 

More recently, another linear discriminant analysis (LDA) model was constructed for a set of 157 compounds for which Pcaco-2 was measured [43]. This model, which applied DRAGON descriptors, achieved an accuracy of classification at 91 % for the training set and 84% for the test set. When this model was applied to predict a set of 241 drugs for which HIA data were available, good correlation (>81%) was achieved between the two ADME-Tox properties. 

Isoniazid derivatives 91 HQSAR and DRAGON descriptors Used HQSAR and generated a test set (R2 = 0.87) for 24 compounds. The results were better than for PLS-QSAR with 2D descriptors from DRAGON (R2 = 0.72) Andrade et al. (34)... 

To overcome these drawbacks of the original formula, an alternative formula JAuthors, This Book], adopted in the —> DRAGON descriptors, is the following ... 

The number of DRAGON descriptors is 1664 in the version 5.4 however, this number increased up to 3224 after inclusion of 2D binary atom pairs and 2D frequency atom pairs, both consisting of 780 descriptors in version 5.5. 

DRAGON descriptors are distinguished into 22 categories, which are listed in Table D8. Table D8 List of descriptor categories in DRAGON software. 

A generalization of the Burden matrix was proposed in DRAGON descriptors where instead of the atomic numbers Z, atomic masses (m), van der Waals volumes (v), Sanderson electronegativities (e), and polarizabilities (p) are used as the weighting schemes for graph vertices. Thus, a general definition of the Burden matrix in terms of a vertex weighting scheme W is... 

Perez Gonzalez, M. and Helguera, A.M. (2003) TOPS-MODE versus DRAGON descriptors to predict permeability coefficients through low-density polyethylene. J. Comput. Aid. Mol. Des., 17, 665-672. 

Perez Gonzalez, M., Helguera, A.M. and Rodriguez, Y.M. (2004) TOPS-MODE and DRAGON descriptors in QSAR. 1. Skin permeation. Internet Electron.]. Mol. Des., 3, 750-758. 

Todeschini, R. (2009) Dragon descriptors. Talete srl, Milan, Italy. 

DRAGON descriptors and the combination MOE/PARASURF show the best correlations between predicted and experimental data. In the second step, the comparison of different machine learning algorithms was performed in combination with descriptors calculated by MOE and PARASURF. It turned out that the predictiv-ity improves by the correction approaches available within Cubist. If the model predictions are corrected based on a kNN approach with five neighbors giving best results, correlation coefficients rise to 0.97 for r (train) and 0.47 for r (test) compared... 

TABLE 11.2 Confusion Matrix for the Cubist Committee Model Based on DRAGON Descriptors... 

Committee models based on DRACjON descriptors or the descriptor combination from MOE and PARASURF appear to result in comparable models with good statistical quality. Since the calculation of PARASURF descriptors employs a semiempirical calculation step, it is significantly more time consuming than the calculation of DRAGON descriptors. Hence, the subsequent classification committee models were built only with DRAGON descriptors. 

For model update, newly collected experimental data were pooled together with the original data, resulting in about 15,000 training and 3,000 test compounds. The final Cubist committee model comprises 602 DRAGON descriptors. Due to the... 

The effect of stmctural fragments on the change of elimination half-life somewhat similar to that described above for the bioavailability. Thus, all lipophilic aromatic fragments had high values (Fig. 14.6). The following model was obtained with Dragon descriptors and MLR method ... 

---