The biological activities of a congeneric series of pyropheophorbides used as sensitizers in photodynamic therapy have been predicted using multiple linear regression (MLR) and artificial neural network (ANN) computations. Theoretical descriptors (81) were calculated by 3DNET program based on the three-dimensional structure of the molecules calculated by MM+ method of Hyperchem program package. These input descriptors were tested as independent variables and used for the model building. The predictive abilities of the best fitting models were checked by leave-one-out cross-validation. ANN was suitable to build models for both linear and non-linear relationships. Lipophilicity was enough to predict accumulation of the sensitizers in the target tissue. Direction dependent eigenvalues like atomic position or mass descriptors, which are perpendicular to the main plane of the molecules, were also needed to predict photodynamic activity. Our models were able to predict biological activities of a series of pyropheophorbide derivatives solely on the basis of their three dimensional molecular structures. ANN turned out to be a valuable method for future model building to predict different biological activities of a wide range of porphyrin based molecules.
|Translated title of the contribution||Prediction of photodynamic activity of photosensitizers using quantitative structureactivity relationship analysis|
|Number of pages||6|
|Journal||Magyar Kemiai Folyoirat, Kemiai Kozlemenyek|
|Publication status||Published - 2001|
ASJC Scopus subject areas