INTERACT-O-FINDER: A Tool for Prediction of DNA-Binding Proteins Using Sequence Features

摘要

Protein-DNA interactions carry out many important regulatory functions in our body which are essential for growth and survival. The molecular level understanding of these interactions helps us to decipher the mechanism of regulation. Conventionally these interactions were identified using small scale experimental techniques and high throughput technologies. But these approaches being time consuming and expensive, arouse the need of a computational approach for the prediction of interactions. To this end, a machine learning approach for predicting and classifying DNA-binding proteins has been used. Weka and LibSVM platform have been used to develop an effective classifying model. The different classifiers were applied on the attributes generated from protein sequence and tried to develop an efficient model for classifying a protein as DNA-interacting or DNA-non-interacting. Among several classifying algorithms applied to generate models, best performance was achieved using LibSVM with 87.83 % accuracy. The tool named INTERACT-O-FINDER, based on the prediction model is available at http://interacto.eurekanow.org/index.html