The Data Mining Approaches for Multi-Class Protein Fold Recognition

Computation analysis of the biological data obtained in genome sequencing and other projects is essential for understanding cellular function and the discovery of new drug and therapies. Data mining become an important tool for researchers of various field including bioinformatics. Protein fold recognition is an important approach to structure discovery in bioinformatics. In this paper the protein fold recognition methods are studied. Supervised learning methods of data mining are carried out and tested for multi-class protein fold recognition. The accuracy is measured by various statistics parameters and the results are reported in this paper. In the result we found that Bayesian Network classifier works better compare as other methods in the cross validation test. The Bayesian Network and Multi Layer Perceptron are reasonably comparable in independent test data supply; accuracy of both methods relatively similar. It is also observed that one-versus-other and all-versus-all mechanisms improve the accuracy as individual parameters.