Identification of defensins employing recurrence quantification analysis and random forest classifiers

Abstract

Defensins represent a class of antimicrobial peptides synthesized in the body acting against various microbes. In this paper we study defensins using a non-linear signal analysis method Recurrence Quantication Analysis (RQA). We used the descriptors calculated employing RQA for the classification of defensins with Random Forest Classifier.The RQA descriptors were able to capture patterns peculiar to defensins leading to an accuracy rate of 78.12% using 10-fold cross validation. © 2009 Springer-Verlag Berlin Heidelberg.