Your browser version is outdated. We recommend that you update your browser to the latest version.

Manuscript Title: Hybrid Learning Algorithm in Neural Network System for Enzyme Classification

Author : Mohd Haniff Osman, Choong-Yeun Liong, and Ishak Hashim

Email : haniff68@ukm.my, lg@ukm.myishak_h@ukm.my

Abstract: Nucleic acid and protein sequences store a wealth of information which ultimately determines their functions and characteristics.Protein sequences classification deals with the assignment of sequences to known categories based on homology detection properties. In this paper, we developed a hybrid learning algorithm in neural network system called Neural Network Enzyme Classification (NNEC) to classify an enzyme found in Protein Data Bank (PDB) to a given family of enzymes. NNEC was developed based on MultilayerPerceptron with hybrid learning algorithm combining the geneticalgorithm (GA) and Backpropagation (BP), where one of them acts as an operator in the other. Here, BP is used as a mutation-like-operator of the general GA search template. The proposed hybrid model was tested with different topologies of network architecture, especially in determining the number of hidden nodes. The precision results are quite promising in classifying the enzyme accordingly.

Keywords: enzyme, protein classification, neural networks, hybrid GA-BP

Vol 2 (2)