Protein solvent accessibility prediction
碩士 === 國立交通大學 === 生物科技研究所 === 91 === To predict protein structures directly from the sequences is one of the most challenging issues in biological research. The knowledge of the solvent accessible surfaces of residues will offer valuable information in predicting the structures and functi...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | en_US |
| Published: |
2003
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/22668187142927917259 |
| Summary: | 碩士 === 國立交通大學 === 生物科技研究所 === 91 === To predict protein structures directly from the sequences is one of the most challenging issues in biological research. The knowledge of the solvent accessible surfaces of residues will offer valuable information in predicting the structures and functions of proteins. The Support Vector Machine (SVM) method has been shown to be quite powerful in solving biological pattern recognition problems. In this study, we will apply SVM methods to the most widely used datasets RS126 and CB421 for solvent accessibility prediction. In our training and testing procedures, we divided the datasets to 20 amino acid groups for training and testing. Our approaches achieve a prediction accuracy 76.1% on CB421 dataset higher than previous SVM approaches. Our results show that SVM using suitable procedures will slightly improve prediction accuracy in solvent accessibility prediction from sequences.
|
|---|