Performance Analysis of Algorithms for Protein Structure Classification

In this paper, we examine the problem of classifying protein fold structure without sequence similarity, by using classification techniques. The representation of the problem in an attribute-based manner allows the application of many well established machine learning algorithms. We study the performance of several algorithms such as decision trees, Naive Bayes, instance-based, and generalized exemplar methods on 27 class problems and on 4 class reduced problems of protein fold classification.