Izvorni znanstveni članak

Quantitative Structure-Activity Relationship of Tricyclic Carbapenems: Application of Artificial Intelligence Methods for Bioactivity Prediction

Mira Lebez ; Faculty of Chemistry and Chemical Technology; University of Ljubljana, Ljubljana, Slovenia
Tom Šolmajer ; Laboratory of Molecular Modeling and NMR Spectroscopy, National Institute of Chemistry, Hajdrihova 19, P. O. Box 660, 1001 Ljubljana, Slovenia
Jure Župan ; Laboratory of Chemometrics, National Institute of Chemistry, Hajdrihova 19, P. O. Box 660, 1001 Ljubljana, Slovenia

Sažetak

Resistance to antibiotics in bacterial population has widened the interest of Scientific community for development of novel therapeutic compounds. Penicillins and cephalosporins which share the β-lactam structural moiety form the most abundant group of antibiotics on the market. Their recently developed tricyclic analogues have shown remarkable bioactivity towards broad spectrum of bacterial species. In a series of 52 tricyclic carbapenems represented by the 180’dimensional »spectrum-like« representation we studied the structure-activity relationships by application of an artificial neural network. The molecular structure representation by spec-tral intensity values served as inputs into the counter-propagation artificial neural network (CP-ANN). SIMPLEX optimization was carried out to obtain the best ANN model and a genetic algorithm approach was subsequently used to simultaneously minimize the number of variables. Thus, a search for the substituents that predominantly influence the experimental bioactivity was performed.
The constructed CP-ANN model yielded bioactivity values predictions with a correlation coefficient of 0.88, with their values extended over 4 orders of magnitude. The list of substituents selected by our automatic procedure can be compared with the data obtained by protein crystallography of the β-lactam inhibitors in complex with D,D-peptidase enzyme.

Ključne riječi

QSAR; tricyclic carbapenem derivatives; antibiotic ac-tivity; articial neural networks; genetic algorithms

Hrčak ID:

127536

URI

https://hrcak.srce.hr/127536

Datum izdavanja:

3.6.2002.

Posjeta: 927 *