| ```markdown | |
| # Goal/Experiment: | |
| Develop a deep learning method for predicting DNA-binding proteins based on contextual features in amino acid sequences. | |
| # An improved deep learning method for predicting DNA-binding proteins based on contextual features in amino acid sequences | |
| *PLOS One* | |
| *Ruixiong Ma* | |
| *USTB* | |
| [DOI Link http://dx.doi.org/10.17504/protocols.io.2rdgd26](http://dx.doi.org/10.17504/protocols.io.2rdgd26) | |
| ## Abstract | |
| With the explosively increased amount of newly discovered proteins, predicting the function of these proteins from amino acid sequences is becoming one of the main challenges in functional annotation of genomes. | |
| Nowadays a number of computational approaches have been developed to predict DNA-binding proteins effectively and accurately from amino acid sequences, such as SVM, DNABP, and CNN-RNN. However, these methods do not consider the context in amino acid sequences, which makes it difficult for them to capture sequence features adequately. | |
| In this paper, we propose CNN-BiLSTM, a new method for predicting DNA-binding proteins, elaborately reconciling convolutional neural network and bi-directional long short-term memory recurrent neural network. CNN-BiLSTM can explore the potential contextual relationships of amino acid sequences to obtain more features than traditional models. | |
| The experimental results show that the prediction accuracy of the proposed CNN-BiLSTM method on the test set is 96.5%, which is 7.8% higher than that of SVM, 9.6% higher than that of DNABP and 3.7% higher than that of CNN-RNN respectively. | |
| Being tested on 20,000 independent samples provided by UniProt that weren't involved in model training, the accuracy of CNN-BiLSTM is 94.5%, which is 12% higher than that of SVM, 4.9% higher than that of DNABP and 4% higher than that of CNN-RNN respectively. | |
| The model training process is visualized and compared with that of CNN-RNN, and it is found that the training process of CNN-BiLSTM support better generalization from the training data set, which shows that CNN-BiLSTM has a wider range of adaptations to protein sequences. | |
| On the independent samples set, CNN-BiLSTM presents better credibility, for its predicted scores are closer to the labels of the samples than those of CNN-RNN. Therefore, the proposed CNN-BiLSTM is a more powerful method for identifying DNA-binding proteins. | |
| ## External Link | |
| [https://doi.org/10.1371/journal.pone.0225317](https://doi.org/10.1371/journal.pone.0225317) | |
| ## Guidelines | |
| This is a method of recognizing DNA binding proteins by deep learning. | |
| ## Materials Text | |
| [Various encrypted data, not transcribed here for brevity] | |
| ## Safety Warnings | |
| Pay attention to the temperature of the computer. | |
| ## Before Starting | |
| ### What you need to prepare: | |
| - Python 3 | |
| - TensorFlow | |
| - Keras | |
| ## Step-by-Step Procedure | |
| ### Step 1: Prepare the dataset | |
| - **Data Extraction**: In the process of extracting data from UniProt, we removed those sequences with length less than 50 or greater than 1,280 amino acids, resulting in 17,651 DNA-binding protein sequences selected as positive samples. At the same time, we got 50,500 non-DNA-binding protein sequences as negative samples. | |
| **UniProt**: UniProt is a comprehensive, high-quality and freely accessible resource of protein sequence and functional information. It provides the scientific community with a vivid, integrated and richly annotated view of protein knowledge. | |
| - **Independent Samples**: Taking sequences from both positive and negative samples as independent test samples, we selected 500 sequences each. | |
| - **Training and Test Sets**: For the remaining 17,151 positive and 50,000 reverse samples, we randomly selected 85% of them as training sets and the remaining 15% as test sets for model training. | |
| ### Step 2: Build Model | |
| The deep learning model is composed of four parts: | |
| 1. **Coding Layer**: Each amino acid is represented as a particular number. | |
| 2. **Embedding Layer**: Translates amino acid sequences into continuous vectors. | |
| 3. **Convolution Layer**: Consists of two convolutions and two maximal pooling operations. | |
| 4. **BiLSTM Layer**: Grasp the context features of amino acid sequences. | |
| We use the Keras platform to build this model. | |
| ### Step 3: Model Training | |
| - The data is trained in the built model using GPU. | |
| - At the end of this process, we get a DNA binding protein predictor. | |
| ## References | |
| This is an open access protocol distributed under the terms of the Creative Commons Attribution License [License URL](https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. | |
| --- | |
| *Note*: This compiled information assists in understanding and applying the described process, ensuring adequate preparation of materials and systematic implementation of the steps. | |
| `endofoutput` | |
| ``` |