Abstract:

Protein-protein interaction sites are very important in drug design and construction of protein interaction networks. A data set that contained 35 proteins was used. First, 3 features of the  proteins were extracted. They were sequence profiles, entropy and accessible surface area. Then back propagation neural networks and their integration were applied to predict proteinprotein interaction sites. One protein was tested each time and the remaining 34 proteins were used to train the classifiers (one-crossover). The final result was good and when adding one new feature the accuracy of prediction  improved. When integrating these features the outcome has more accuracy than a single  classifier.

Key words:  protein-protein interaction sites, sequence profiles, entropy, accessible surface area, ensemble neural networks