Coevolution based prediction of protein-protein interactions with reduced training data

Protein-protein interactions are important for the prediction of protein functions since two interacting proteins usually have similar functions in a cell. In this work, our aim is to predict protein-protein interactions with a known portion of the interaction network when there are large numbers of protein interactions in the data set. Phylogenetic profiles of proteins form the feature vectors for training Support Vector Machine (SVM). To reduce the training time of SVM we reduced the data size by k-means and MEB clustering techniques and we applied feature selection methods by selecting most representative features by phylogenetic tree and Fisher´s Exact Test methods. The training data clustered by the k-means method gave superior results in prediction accuracies.