Algorithm for Predicting Functionally Equivalent Proteins from BLAST and HMMER Searches

作者：Yu Dong Su; Lee Dae Hee; Kim Seong Keun; Lee Choong Hoon; Song Ju Yeon; Kong Eun Bae^*; Kim Jihyun F

来源：Journal of Microbiology and Biotechnology, 2012, 22(8): 1054-1058.

DOI：10.4014/jmb.1203.03050

• Functionally equivalent protein
• error backpropagation algorithm
• sequence-based method
• artificial neural network
• bioinformatics

In order to predict biologically significant attributes such as function from protein sequences, searching against large databases for homologous proteins is a common practice. In particular, BLAST and HMMER are widely used in a variety of biological fields. However, sequence-homologous proteins determined by BLAST and proteins having the same domains predicted by HMMER are not always functionally equivalent, even though their sequences are aligning with high similarity. Thus, accurate assignment of functionally equivalent proteins from aligned sequences remains a challenge in bioinformatics. We have developed the FEP-BH algorithm to predict functionally equivalent proteins from protein protein pairs identified by BLAST and from protein domain pairs predicted by HMMER. When examined against domain classes of the Pfam-A seed database, FEP-BH showed 71.53% accuracy, whereas BLAST and HMMER were 57.72% and 36.62%, respectively. We expect that the FEP-BH algorithm will be effective in predicting functionally equivalent proteins from BLAST and HMMER outputs and will also suit biologists who want to search out functionally equivalent proteins from among sequence-homologous proteins.

• 出版日期2012-8