Classification of Enzyme Function from Protein Sequence based on Feature Representation

• Main Authors: Lee, Bum Ju, Lee, Heon Gu, Lee, Jong Yun, Ryu, Keun Ho
• Format: Conference Proceeding
• Language: English
• Subjects: amino acid composition; Amino acids; attribute selection; Biochemistry; Bioinformatics; Chemical analysis; enzyme function; feature analysis; Feature extraction; function classification; Laboratories; machine learning; protein classification; Protein engineering; Protein sequence; Sequences; Spatial databases
• DOI: 10.1109/BIBE.2007.4375643

Enzymes are the proteins that accelerate the rate of chemical reaction, and both their structures and dynamics may be important to their function of catalyzing biochemical reactions. For the function prediction and classification of enzymes, many methods based on sequence similarity to detect similar proteins have been developed. However, these methods often miscarry in the case of the absence of similar sequences or poor similarity among proteins. Therefore, many researchers have been developing alternative approaches that assign function from protein features without consideration of sequence similarity. In this paper, we propose a method of sequence-driven feature extraction and enzyme functional classification using only the features of protein sequence, excluding predicted secondary structures and annotation information of protein databases. Our experimental results demonstrate that the enzyme classification based on the Chi-Squared ranking method among various attribute selection methods is efficient. Also, we And that amino acid composition of specific enzyme differs from composition of other enzymes.