کد مقاله | کد نشریه | سال انتشار | مقاله انگلیسی | نسخه تمام متن |
---|---|---|---|---|
4495981 | 1623826 | 2015 | 6 صفحه PDF | دانلود رایگان |
• Computational model is developed for prediction of membrane protein types.
• PseAAC is used as feature extraction scheme.
• Various classification algorithms are utilized.
• VFI achieved quite promising results.
Membrane protein is a major constituent of cell, performing numerous crucial functions in the cell. These functions are mostly concerned with membrane protein׳s types. Initially, membrane proteins types are classified through traditional methods and reasonable results were obtained using these methods. However, due to large exploration of protein sequences in databases, it is very difficult or sometimes impossible to classify through conventional methods, because it is laborious and wasting of time. Therefore, a new powerful discriminating model is indispensable for classification of membrane protein׳s types with high precision. In this work, a quite promising classification model is developed having effective discriminating power of membrane protein׳s types. In our classification model, silent features of protein sequences are extracted via Pseudo Amino Acid Composition. Five classification algorithms were utilized. Among these classification algorithms Voting Feature Interval has obtained outstanding performance in all the three datasets. The accuracy of proposed model is 93.9% on dataset S1, 89.33% on S2 and 86.9% on dataset S3, respectively, applying 10-fold cross validation test. The success rates revealed that our proposed model has obtained the utmost outcomes than other existing models in literatures so far and will be played a substantial role in the fields of drug design and pharmaceutical industry.
Journal: Journal of Theoretical Biology - Volume 384, 7 November 2015, Pages 78–83