scholarly journals Protein Secondary Structure Prediction using Recurrent Neural Networks

2020 ◽  
Vol 8 (5) ◽  
pp. 660-663
Keyword(s):  
Neural Networks ◽  
Amino Acid ◽  
Secondary Structure ◽  
Amino Acid Sequence ◽  
Recurrent Neural Networks ◽  
Protein Sequence ◽  
Structure Prediction ◽  
Short Term Memory ◽  
Secondary Structure Prediction ◽  
Protein Secondary Structure

In bioinformatics the prediction of the secondary structure of the protein from its primary amino acid sequence is very difficult, which has a huge impact on the field of science and medicine. The hardest part is how to learn the most effective and correct protein features to improve prediction. Here, we carry out a deep learning model to enhance structure prediction. The core achievement of this paper is a group of recurrent neural networks (RNNs) that can manage high-level relational features from a pair of input protein sequence and target protein sequences. This paper contrasts the different type of recurrent network in recurrent neural networks (RNNs). In addition, the emphasis is on more advanced systems which incorporate a gating utility is called long short term memory (LSTM) unit and the newly added gated recurrent unit (GRU). This recurrent units has been calculated on the basis of predicting protein secondary structure using an amino acid sequence. The dataset has been taken from a publicly available database server (RCSB), and this study shows that advanced recurrent units LSTM is better than GRU for a long protein sequence.

Protein secondary structure prediction improved by recurrent neural networks integrated with two-dimensional convolutional neural networks

2018 ◽  
Vol 16 (05) ◽  
pp. 1850021 ◽  
Author(s):  
Yanbu Guo ◽  
Bingyi Wang ◽  
Weihua Li ◽  
Bei Yang
Keyword(s):  
Neural Networks ◽  
Amino Acid ◽  
Secondary Structure ◽  
Structure Prediction ◽  
Feature Vector ◽  
Secondary Structure Prediction ◽  
Protein Secondary Structure ◽  
Two Dimensional ◽  
The Matrix ◽  
Vector Dimension

Protein secondary structure prediction (PSSP) is an important research field in bioinformatics. The representation of protein sequence features could be treated as a matrix, which includes the amino-acid residue (time-step) dimension and the feature vector dimension. Common approaches to predict secondary structures only focus on the amino-acid residue dimension. However, the feature vector dimension may also contain useful information for PSSP. To integrate the information on both dimensions of the matrix, we propose a hybrid deep learning framework, two-dimensional convolutional bidirectional recurrent neural network (2C-BRNN), for improving the accuracy of 8-class secondary structure prediction. The proposed hybrid framework is to extract the discriminative local interactions between amino-acid residues by two-dimensional convolutional neural networks (2DCNNs), and then further capture long-range interactions between amino-acid residues by bidirectional gated recurrent units (BGRUs) or bidirectional long short-term memory (BLSTM). Specifically, our proposed 2C-BRNNs framework consists of four models: 2DConv-BGRUs, 2DCNN-BGRUs, 2DConv-BLSTM and 2DCNN-BLSTM. Among these four models, the 2DConv- models only contain two-dimensional (2D) convolution operations. Moreover, the 2DCNN- models contain 2D convolutional and pooling operations. Experiments are conducted on four public datasets. The experimental results show that our proposed 2DConv-BLSTM model performs significantly better than the benchmark models. Furthermore, the experiments also demonstrate that the proposed models can extract more meaningful features from the matrix of proteins, and the feature vector dimension is also useful for PSSP. The codes and datasets of our proposed methods are available at https://github.com/guoyanb/JBCB2018/ .

Sign in / Sign up

Export Citation Format

Share Document