engleski

Simplified structure-solubility relationships for early ADME evaluation in drug discovery

Relationship between molecular structure and solubility of 1039 organic compounds were studied. Initial structures of molecules were encoded as SMILES and converted to the 3D structures by the CORINA program (www2.chemie.uni-erlangen.de/software/corina/) and more than 1000 initial descriptors were computed by the program Dragon 2.1 (http://www.disat.unimib.it/chm/). Initial set of descriptors was filtered in order to remove non-significant and highly inter-correlated descriptors (123 descriptors remained after filtering). Finally, the best linear multiregression models containing 1-7 descriptors were selected from the set of 123 descriptors by the CROMRsel program (B. Lučić ; N. Trinajstić, J. Chem. Inf. Comput. Sci. 1999, 39, 121-132). Standard error of estimate and standard error of leave-one-out cross-validation obtained on the training set are 0.739 and 0.746 log units, respectively. In the best seven-descriptor model five topological descriptors, one atom-centered fragment descriptor and calculated Moriguchi octanol-water partition coefficient are involved. Using the best linear seven-descriptor model we performed prediction on external data set containing 258 molecules. We obtained the same prediction of solubility for 258 molecules from the test set (Stst = 0.745 log units), as it was obtained by the neural network model developed on the same sets of molecules (R. Liu and S.-S. So, J. Chem. Inf. Comput. Sci. 2001, 41, 1633). It should be mentioned that the statistical parameter (Stst = 0.71 log units) given in Table 1 (page 1636) for the best neural network model was not computed correctly - correct value is Stst = 0.745 log units. However, our seven-descriptor model is linear (containing only eight optimized parameters) and much simpler than the above mentioned neural network model (which is a nonlinear model, containing seven descriptors and having 19 optimized parameters). Presented linear multiregression models can be additionally improved by the inclusion of nonlinear terms of initial descriptors.

nije evidentirano

nije evidentirano