Shun Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Shun Zhang has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Chun Fang
College of Information Engineering, Beijing Institute of Petrochemical Technology, Beijing, China
4 shared publications - Jiasheng He
Department of Information Engineering, Beijing Institute of Petrochemical Technology, 19 Qingyuan North Road, Daxing District, Beijing 102617, P. R. China
2 shared publications - Fan Zhang
Beijing Institute of Petrochemical Technology, Beijing, 102617, China
1 shared publication - Jinfeng Li
Beijing Institute of Petrochemical Technology, China
1 shared publication
Latest Publications
- Open Access
- 26 January 2024
- Published by Association for Computing Machinery (ACM)
- p. 375-379
Prediction of DNA enhancers based on multi-species genomic base model DNABERT-2 and BiGRU network
DNA enhancers are important regulatory elements in the genome, and accurate prediction is crucial for understanding gene regulation. In this study, a DNA enhancer prediction algorithm, DNABERT-Enhancer, is proposed by combining the DNABERT-2 model and the BiGRU network. The algorithm analogizes DNA sequences to sentences, and strengthens the capture of contextual information of DNA sequences by integrating the structural strengths of the DNABERT-2 pre-training model and the BiGRU network, thus realizing the DNA enhancers prediction. The experimental results show an AUC value of 0.943 on the test set. This study provides a new method for DNA enhancer prediction in the field of bioinformatics, which is expected to enhance our understanding of gene regulation in practical applications, for example, it may be used to more accurately identify disease markers, customize personalized medical treatment plans or finding new targets in drug discovery that advances the field of genomics.
- 26 January 2024
- Published by Association for Computing Machinery (ACM)
- p. 375-379
Open AccessPrediction of DNA enhancers based on multi-species genomic base model DNABERT-2 and BiGRU network
DNA enhancers are important regulatory elements in the genome, and accurate prediction is crucial for understanding gene regulation. In this study, a DNA enhancer prediction algorithm, DNABERT-Enhancer, is proposed by combining the DNABERT-2 model and the BiGRU network. The algorithm analogizes DNA sequences to sentences, and strengthens the capture of contextual information of DNA sequences by integrating the structural strengths of the DNABERT-2 pre-training model and the BiGRU network, thus realizing the DNA enhancers prediction. The experimental results show an AUC value of 0.943 on the test set. This study provides a new method for DNA enhancer prediction in the field of bioinformatics, which is expected to enhance our understanding of gene regulation in practical applications, for example, it may be used to more accurately identify disease markers, customize personalized medical treatment plans or finding new targets in drug discovery that advances the field of genomics.
- 28 October 2023
- Published by World Scientific Pub Co Pte Ltd in Journal of Bioinformatics and Computational Biology
- Vol. 21 (05) , 2350022
AAindex-PPII: Predicting polyproline type II helix structure based on amino acid indexes with an improved BiGRU-TextCNN model
The polyproline-II (PPII) structure domain is crucial in organisms’ signal transduction, transcription, cell metabolism, and immune response. It is also a critical structural domain for specific vital disease-associated proteins. Recognizing PPII is essential for understanding protein structure and function. To accurately predict PPII in proteins, we propose a novel method, AAindex-PPII, which only adopts amino acid index to characterize protein sequences and uses a Bidirectional Gated Recurrent Unit (BiGRU)-Improved TextCNN composite deep learning model to predict PPII in proteins. Experimental results show that, when tested on the same datasets, our method outperforms the state-of-the-art BERT-PPII method, achieving an AUC value of 0.845 on the strict data and an AUC value of 0.813 on the non-strict data, which is 0.024 and 0.03 higher than that of the BERT-PPII method. This study demonstrates that our proposed method is simple and efficient for PPII prediction without using pre-trained large models or complex features such as position-specific scoring matrices.
- 28 October 2023
- Published by World Scientific Pub Co Pte Ltd in Journal of Bioinformatics and Computational Biology
- Vol. 21 (05) , 2350022
AAindex-PPII: Predicting polyproline type II helix structure based on amino acid indexes with an improved BiGRU-TextCNN model
The polyproline-II (PPII) structure domain is crucial in organisms’ signal transduction, transcription, cell metabolism, and immune response. It is also a critical structural domain for specific vital disease-associated proteins. Recognizing PPII is essential for understanding protein structure and function. To accurately predict PPII in proteins, we propose a novel method, AAindex-PPII, which only adopts amino acid index to characterize protein sequences and uses a Bidirectional Gated Recurrent Unit (BiGRU)-Improved TextCNN composite deep learning model to predict PPII in proteins. Experimental results show that, when tested on the same datasets, our method outperforms the state-of-the-art BERT-PPII method, achieving an AUC value of 0.845 on the strict data and an AUC value of 0.813 on the non-strict data, which is 0.024 and 0.03 higher than that of the BERT-PPII method. This study demonstrates that our proposed method is simple and efficient for PPII prediction without using pre-trained large models or complex features such as position-specific scoring matrices.
- Open Access
- 22 September 2023
- Published by Association for Computing Machinery (ACM)
- p. 17-21
CPPGAN-ESM: GAN-Enhanced ESM Prediction Model for Cell-Penetrating Peptide
Cell-penetrating peptide, with their ability to traverse cell membranes and access the intracellular environment, has garnered significant attention in contemporary medical research. Predicting cell-penetrating peptide can expand the therapeutic scope, improve the therapeutic effect, and promote the development of drug delivery, gene therapy, and other fields. Compared with the traditional wet-lab method, predicting cell-penetrating peptide by computational methods has the advantages of low cost and accuracy. However, due to the difficulty of collecting cell-penetrating peptide datasets and the performance limitation of traditional feature extractors, Predicting the performance of cell-penetrating peptide using traditional computational methods has limitations. To address these problems, a cell-penetrating peptide prediction method named CPPGAN-ESM is proposed in this paper. CPPGAN-ESM firstly augments the cell-penetrating peptide dataset by using Deep Convolutional Generative Adversarial Network, and then fine-tunes the predictor based on the Evolutionary Scale Modeling 2 pre-trained feature extractor with the augmented dataset, to realize the prediction of the cell-penetrating peptide. Extensive experiments on a public dataset show that the method achieves an efficient performance of 0.979 AUC and 0.977 ACC in the cell-penetrating peptide prediction task.
- 22 September 2023
- Published by Association for Computing Machinery (ACM)
- p. 17-21
Open AccessCPPGAN-ESM: GAN-Enhanced ESM Prediction Model for Cell-Penetrating Peptide
Cell-penetrating peptide, with their ability to traverse cell membranes and access the intracellular environment, has garnered significant attention in contemporary medical research. Predicting cell-penetrating peptide can expand the therapeutic scope, improve the therapeutic effect, and promote the development of drug delivery, gene therapy, and other fields. Compared with the traditional wet-lab method, predicting cell-penetrating peptide by computational methods has the advantages of low cost and accuracy. However, due to the difficulty of collecting cell-penetrating peptide datasets and the performance limitation of traditional feature extractors, Predicting the performance of cell-penetrating peptide using traditional computational methods has limitations. To address these problems, a cell-penetrating peptide prediction method named CPPGAN-ESM is proposed in this paper. CPPGAN-ESM firstly augments the cell-penetrating peptide dataset by using Deep Convolutional Generative Adversarial Network, and then fine-tunes the predictor based on the Evolutionary Scale Modeling 2 pre-trained feature extractor with the augmented dataset, to realize the prediction of the cell-penetrating peptide. Extensive experiments on a public dataset show that the method achieves an efficient performance of 0.979 AUC and 0.977 ACC in the cell-penetrating peptide prediction task.
- Open Access
- 22 September 2023
- Published by Association for Computing Machinery (ACM)
- p. 242-246
PB_DBP: Identifying DNA-binding proteins using ProBert_BiLSTM model
DNA-binding proteins are a class of proteins that can bind to specific nucleotide sequences in the DNA sequence, and they play essential roles in maintaining the structure and function of DNA, regulating the expression of genes, and participating in the cell cycle. Traditional prediction methods usually focus on extracting the physicochemical characteristics of proteins from sequences, ignoring the information inherent in amino acid sequences and the positional information between amino acid sequences. Moreover, the accuracy and reliability of the prediction are limited by the small size and noise of the training data sets. This study introduces a new prediction method called PB_DBP to tackle the issues mentioned above. PB_DBP combines ProtBert, a pre-training model designed for protein sequences, with BiLSTM, a model capable of effectively capturing long-term dependencies in sequence data. Our method has been demonstrated to outperform the combination of CNN and LSTM on the public data sets, achieving an AUC value of 0.988.
- 22 September 2023
- Published by Association for Computing Machinery (ACM)
- p. 242-246
Open AccessPB_DBP: Identifying DNA-binding proteins using ProBert_BiLSTM model
DNA-binding proteins are a class of proteins that can bind to specific nucleotide sequences in the DNA sequence, and they play essential roles in maintaining the structure and function of DNA, regulating the expression of genes, and participating in the cell cycle. Traditional prediction methods usually focus on extracting the physicochemical characteristics of proteins from sequences, ignoring the information inherent in amino acid sequences and the positional information between amino acid sequences. Moreover, the accuracy and reliability of the prediction are limited by the small size and noise of the training data sets. This study introduces a new prediction method called PB_DBP to tackle the issues mentioned above. PB_DBP combines ProtBert, a pre-training model designed for protein sequences, with BiLSTM, a model capable of effectively capturing long-term dependencies in sequence data. Our method has been demonstrated to outperform the combination of CNN and LSTM on the public data sets, achieving an AUC value of 0.988.