Improving SIM-based annotation method of protein sequence using support vector machine

• Main Authors: Cheng-Kang Liu, 劉承剛
• Other Authors: Hahn-Ming Lee
• Format: Others
• Language: en_US
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/05841332477652714545

碩士 === 國立臺灣科技大學 === 資訊工程系 === 93 === 　The gap between the protein sequences and the reliable function annotation in public databases is growing. Traditional manual annotation by literature curation can not catch up with the rapid growth of new protein sequences. Thus, the automatic annotation methods of protein sequences are in great demand are in great demand. Sequence similarity (SIM) methods, such as BLAST, are the most common used method which searching for homologies and evolutionary relationship between the protein sequences. However, there are a considerable number of functional inconsistencies in similar protein sequences. Thus, a method to automatic eliminates the error annotations is needed to improve the SIM-based methods. In addition, the biological data are distributed in different databases and having their own data types. It is difficult for users to obtain these data they needed from the distributed environment. Integration of the various types of biological data into an integrated environment for function annotation of protein sequences is also an important issue. In this paper, we present a protein sequence annotation method, named as MAPS (Multiple Annotation for Protein Sequences), which provides a mechanism to extract multiple annotations from various types of biological data and automatic eliminates the error annotations by a pre-trained SVM classifier. It assigns an annotation to the input protein sequence by taking into account all hit proteins with this annotation entirely, not only single hit protein. This can reduce the error annotations inferred from weak sequence similarity and the sequences identity in non-functional segment. The experimental results show that the error annotations can be eliminated effectively and keep high accuracy on different types of annotations.