Using Deep Learning (CNN, RNN, LSTM, GRU) methods for the prediction of Protein Secondary Structure
Proteins play a crucial function in the biological processes of living organisms. Knowing the function of the protein offers significant insight into future biological and medical research. Since a protein’s shape determines its function, it is important to understand the protein’s 3D structure. Alt...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Istanbul University Press
2022-06-01
|
| Series: | Acta Infologica |
| Subjects: | |
| Online Access: | https://cdn.istanbul.edu.tr/file/JTA6CLJ8T5/8C36525B2D2941E68DD9A77636179D9A |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Proteins play a crucial function in the biological processes of living organisms. Knowing the function of the protein offers significant insight into future biological and medical research. Since a protein’s shape determines its function, it is important to understand the protein’s 3D structure. Although experimental methods such as X-ray crystallography and nuclear magnetic resonance (NMR) have been used to examine the shape of proteins, so far the results have been insufficient. As a result, predicting the 3D structure of proteins is crucial. Determining the 3D structure of a protein from its primary structure is challenging. Therefore, predicting the protein secondary structure becomes important for studying its structure and function. Many emerging methods, including machine learning, as well as deep learning, have been used to predict the secondary structure of proteins and comprise a crucial part of Structural Bioinformatics. The goal of this study is to compare the results generated by predictive models that were created using the four most frequently utilized deep learning methods: convolutional neural networks (CNN), recurrent neural networks (RNN), long short term memory networks (LSTM), and gated recurrent units (GRU). The CB513 dataset was used to train and test these models, and performance evaluation metrics viz. accuracy, f1 score, recall, and precision were applied. The CNN, RNN, LSTM, and GRU models had an accuracy of 82.54%, 82.06%, 81.1%, and 81.48%, respectively. |
|---|---|
| ISSN: | 2602-3563 |