1
|
Naseem A, Khan YD. An intelligent model for prediction of abiotic stress-responsive microRNAs in plants using statistical moments based features and ensemble approaches. Methods 2024; 228:65-79. [PMID: 38768931 DOI: 10.1016/j.ymeth.2024.05.008] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 04/04/2024] [Revised: 04/30/2024] [Accepted: 05/10/2024] [Indexed: 05/22/2024] Open
Abstract
This study proposed an intelligent model for predicting abiotic stress-responsive microRNAs in plants. MicroRNAs (miRNAs) are short RNA molecules regulates the stress in genes. Experimental methods are costly and time-consuming, as compare to in-silico prediction. Addressing this gap, the study seeks to develop an efficient computational model for plant stress response prediction. The two benchmark datasets for MiRNA and Pre-MiRNA dataset have been acquired in this study. Four ensemble approaches such as bagging, boosting, stacking, and blending have been employed. Classifiers such as Random Forest (RF), Extra Trees (ET), Ada Boost (ADB), Light Gradient Boosting Machine (LGBM), and Support Vector Machine (SVM). Stacking and Blending employed all stated classifiers as base learners and Logistic Regression (LR) as Meta Classifier. There have been a total of four types of testing used, including independent set, self-consistency, cross-validation with 5 and 10 folds, and jackknife. This study has utilized evaluation metrics such as accuracy score, specificity, sensitivity, Mathew's correlation coefficient (MCC), and AUC. Our proposed methodology has outperformed existing state of the art study in both datasets based on independent set testing. The SVM-based approach has exhibited accuracy score of 0.659 for the MiRNA dataset, which is better than the previous study. The ET classifier has surpassed the accuracy of Pre-MiRNA dataset as compared to the existing benchmark study, achieving an impressive score of 0.67. The proposed method can be used in future research to predict abiotic stresses in plants.
Collapse
Affiliation(s)
- Ansar Naseem
- Department of Artificial Intelligence, School of Systems and Technology, University of Management and Technology, Lahore, Pakistan
| | - Yaser Daanial Khan
- Department of Computer Science, School of Systems and Technology, University of Management and Technology, Lahore, Pakistan.
| |
Collapse
|
2
|
Malebary SJ, Alromema N, Suleman MT, Saleem M. m5c-iDeep: 5-Methylcytosine sites identification through deep learning. Methods 2024; 230:80-90. [PMID: 39089345 DOI: 10.1016/j.ymeth.2024.07.008] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/09/2024] [Revised: 07/16/2024] [Accepted: 07/23/2024] [Indexed: 08/03/2024] Open
Abstract
5-Methylcytosine (m5c) is a modified cytosine base which is formed as the result of addition of methyl group added at position 5 of carbon. This modification is one of the most common PTM that used to occur in almost all types of RNA. The conventional laboratory methods do not provide quick reliable identification of m5c sites. However, the sequence data readiness has made it feasible to develop computationally intelligent models that optimize the identification process for accuracy and robustness. The present research focused on the development of in-silico methods built using deep learning models. The encoded data was then fed into deep learning models, which included gated recurrent unit (GRU), long short-term memory (LSTM), and bi-directional LSTM (Bi-LSTM). After that, the models were subjected to a rigorous evaluation process that included both independent set testing and 10-fold cross validation. The results revealed that LSTM-based model, m5c-iDeep, outperformed revealing 99.9 % accuracy while comparing with existing m5c predictors. In order to facilitate researchers, m5c-iDeep was also deployed on a web-based server which is accessible at https://taseersuleman-m5c-ideep-m5c-ideep.streamlit.app/.
Collapse
Affiliation(s)
- Sharaf J Malebary
- Department of Information Technology, Faculty of Computing and Information Technology-Rabigh, King Abdulaziz University, P.O. Box 344, Rabigh 21911, Saudi Arabia
| | - Nashwan Alromema
- Department of Computer Science, Faculty of Computing and Information Technology-Rabigh, King Abdulaziz University, P.O. Box 344, Rabigh 21911, Saudi Arabia.
| | - Muhammad Taseer Suleman
- Department of Criminology and Forensic Sciences, Lahore Garrison University, Lahore Pakistan; Department of Computer Science, School of Systems and Technology, University of Management and Technology, Lahore 54770 Pakistan
| | - Maham Saleem
- Department of Computer Science, School of Systems and Technology, University of Management and Technology, Lahore 54770 Pakistan
| |
Collapse
|
3
|
Pradhan UK, Meher PK, Naha S, Sharma NK, Agarwal A, Gupta A, Parsad R. DBPMod: a supervised learning model for computational recognition of DNA-binding proteins in model organisms. Brief Funct Genomics 2024; 23:363-372. [PMID: 37651627 DOI: 10.1093/bfgp/elad039] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/08/2023] [Revised: 08/09/2023] [Accepted: 08/15/2023] [Indexed: 09/02/2023] Open
Abstract
DNA-binding proteins (DBPs) play critical roles in many biological processes, including gene expression, DNA replication, recombination and repair. Understanding the molecular mechanisms underlying these processes depends on the precise identification of DBPs. In recent times, several computational methods have been developed to identify DBPs. However, because of the generic nature of the models, these models are unable to identify species-specific DBPs with higher accuracy. Therefore, a species-specific computational model is needed to predict species-specific DBPs. In this paper, we introduce the computational DBPMod method, which makes use of a machine learning approach to identify species-specific DBPs. For prediction, both shallow learning algorithms and deep learning models were used, with shallow learning models achieving higher accuracy. Additionally, the evolutionary features outperformed sequence-derived features in terms of accuracy. Five model organisms, including Caenorhabditis elegans, Drosophila melanogaster, Escherichia coli, Homo sapiens and Mus musculus, were used to assess the performance of DBPMod. Five-fold cross-validation and independent test set analyses were used to evaluate the prediction accuracy in terms of area under receiver operating characteristic curve (auROC) and area under precision-recall curve (auPRC), which was found to be ~89-92% and ~89-95%, respectively. The comparative results demonstrate that the DBPMod outperforms 12 current state-of-the-art computational approaches in identifying the DBPs for all five model organisms. We further developed the web server of DBPMod to make it easier for researchers to detect DBPs and is publicly available at https://iasri-sg.icar.gov.in/dbpmod/. DBPMod is expected to be an invaluable tool for discovering DBPs, supplementing the current experimental and computational methods.
Collapse
Affiliation(s)
- Upendra K Pradhan
- Division of Statistical Genetics, ICAR-Indian Agricultural Statistics Research Institute, PUSA, New Delhi 110012, India
| | - Prabina K Meher
- Division of Statistical Genetics, ICAR-Indian Agricultural Statistics Research Institute, PUSA, New Delhi 110012, India
| | - Sanchita Naha
- Division of Computer Applications, ICAR-Indian Agricultural Statistics Research Institute, PUSA, New Delhi 110012, India
| | - Nitesh K Sharma
- Titus Family Department of Clinical Pharmacy, USC Alfred E. Mann School of Pharmacy and Pharmaceutical Sciences, University of Southern California, 1540 Alcazar Street, Los Angeles, CA 90033, USA
| | - Aarushi Agarwal
- Amity Institute of Biotechnology, Amity University, Noida, Uttar Pradesh 201313, India
| | - Ajit Gupta
- Division of Statistical Genetics, ICAR-Indian Agricultural Statistics Research Institute, PUSA, New Delhi 110012, India
| | - Rajender Parsad
- ICAR-Indian Agricultural Statistics Research Institute, PUSA, New Delhi 110012, India
| |
Collapse
|
4
|
Pradhan UK, Meher PK, Naha S, Das R, Gupta A, Parsad R. ProkDBP: Toward more precise identification of prokaryotic DNA binding proteins. Protein Sci 2024; 33:e5015. [PMID: 38747369 PMCID: PMC11094783 DOI: 10.1002/pro.5015] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/08/2023] [Revised: 04/18/2024] [Accepted: 04/21/2024] [Indexed: 05/19/2024]
Abstract
Prokaryotic DNA binding proteins (DBPs) play pivotal roles in governing gene regulation, DNA replication, and various cellular functions. Accurate computational models for predicting prokaryotic DBPs hold immense promise in accelerating the discovery of novel proteins, fostering a deeper understanding of prokaryotic biology, and facilitating the development of therapeutics targeting for potential disease interventions. However, existing generic prediction models often exhibit lower accuracy in predicting prokaryotic DBPs. To address this gap, we introduce ProkDBP, a novel machine learning-driven computational model for prediction of prokaryotic DBPs. For prediction, a total of nine shallow learning algorithms and five deep learning models were utilized, with the shallow learning models demonstrating higher performance metrics compared to their deep learning counterparts. The light gradient boosting machine (LGBM), coupled with evolutionarily significant features selected via random forest variable importance measure (RF-VIM) yielded the highest five-fold cross-validation accuracy. The model achieved the highest auROC (0.9534) and auPRC (0.9575) among the 14 machine learning models evaluated. Additionally, ProkDBP demonstrated substantial performance with an independent dataset, exhibiting higher values of auROC (0.9332) and auPRC (0.9371). Notably, when benchmarked against several cutting-edge existing models, ProkDBP showcased superior predictive accuracy. Furthermore, to promote accessibility and usability, ProkDBP (https://iasri-sg.icar.gov.in/prokdbp/) is available as an online prediction tool, enabling free access to interested users. This tool stands as a significant contribution, enhancing the repertoire of resources for accurate and efficient prediction of prokaryotic DBPs.
Collapse
Affiliation(s)
- Upendra Kumar Pradhan
- Division of Statistical GeneticsICAR‐Indian Agricultural Statistics Research Institute, PUSANew DelhiIndia
| | - Prabina Kumar Meher
- Division of Statistical GeneticsICAR‐Indian Agricultural Statistics Research Institute, PUSANew DelhiIndia
| | - Sanchita Naha
- Division of Computer ApplicationsICAR‐Indian Agricultural Statistics Research Institute, PUSANew DelhiIndia
| | - Ritwika Das
- Division of Agricultural BioinformaticsICAR‐Indian Agricultural Statistics Research Institute, PUSANew DelhiIndia
| | - Ajit Gupta
- Division of Statistical GeneticsICAR‐Indian Agricultural Statistics Research Institute, PUSANew DelhiIndia
| | - Rajender Parsad
- ICAR‐Indian Agricultural Statistics Research Institute, PUSANew DelhiIndia
| |
Collapse
|
5
|
Suleman MT, Alturise F, Alkhalifah T, Khan YD. m1A-Ensem: accurate identification of 1-methyladenosine sites through ensemble models. BioData Min 2024; 17:4. [PMID: 38360720 PMCID: PMC10868122 DOI: 10.1186/s13040-023-00353-x] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/30/2023] [Accepted: 12/31/2023] [Indexed: 02/17/2024] Open
Abstract
BACKGROUND 1-methyladenosine (m1A) is a variant of methyladenosine that holds a methyl substituent in the 1st position having a prominent role in RNA stability and human metabolites. OBJECTIVE Traditional approaches, such as mass spectrometry and site-directed mutagenesis, proved to be time-consuming and complicated. METHODOLOGY The present research focused on the identification of m1A sites within RNA sequences using novel feature development mechanisms. The obtained features were used to train the ensemble models, including blending, boosting, and bagging. Independent testing and k-fold cross validation were then performed on the trained ensemble models. RESULTS The proposed model outperformed the preexisting predictors and revealed optimized scores based on major accuracy metrics. CONCLUSION For research purpose, a user-friendly webserver of the proposed model can be accessed through https://taseersuleman-m1a-ensem1.streamlit.app/ .
Collapse
Affiliation(s)
- Muhammad Taseer Suleman
- Department of Computer Science, School of Systems and Technology, University of Management and Technology, Lahore, 54770, Pakistan
| | - Fahad Alturise
- Department of Computer, College of Science and Arts in Ar Rass, Qassim University, Ar Rass, Qassim, Saudi Arabia.
| | - Tamim Alkhalifah
- Department of Computer, College of Science and Arts in Ar Rass, Qassim University, Ar Rass, Qassim, Saudi Arabia
| | - Yaser Daanial Khan
- Department of Computer Science, School of Systems and Technology, University of Management and Technology, Lahore, 54770, Pakistan
| |
Collapse
|
6
|
Naseem A, Alturise F, Alkhalifah T, Khan YD. BBB-PEP-prediction: improved computational model for identification of blood-brain barrier peptides using blending position relative composition specific features and ensemble modeling. J Cheminform 2023; 15:110. [PMID: 37980534 PMCID: PMC10656963 DOI: 10.1186/s13321-023-00773-1] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/07/2023] [Accepted: 10/21/2023] [Indexed: 11/20/2023] Open
Abstract
BBPs have the potential to facilitate the delivery of drugs to the brain, opening up new avenues for the development of treatments targeting diseases of the central nervous system (CNS). The obstacle faced in central nervous system disorders stems from the formidable task of traversing the blood-brain barrier (BBB) for pharmaceutical agents. Nearly 98% of small molecule-based drugs and nearly 100% of large molecule-based drugs encounter difficulties in successfully penetrating the BBB. This importance leads to identification of these peptides, can help in healthcare systems. In this study, we proposed an improved intelligent computational model BBB-PEP-Prediction for identification of BBB peptides. Position and statistical moments based features have been computed for acquired benchmark dataset. Four types of ensembles such as bagging, boosting, stacking and blending have been utilized in the methodology section. Bagging employed Random Forest (RF) and Extra Trees (ET), Boosting utilizes XGBoost (XGB) and Light Gradient Boosting Machine (LGBM). Stacking uses ET and XGB as base learners, blending exploited LGBM and RF as base learners, while Logistic Regression (LR) has been applied as Meta learner for stacking and blending. Three classifiers such as LGBM, XGB and ET have been optimized by using Randomized search CV. Four types of testing such as self-consistency, independent set, cross-validation with 5 and 10 folds and jackknife test have been employed. Evaluation metrics such as Accuracy (ACC), Specificity (SPE), Sensitivity (SEN), Mathew's correlation coefficient (MCC) have been utilized. The stacking of classifiers has shown best results in almost each testing. The stacking results for independent set testing exhibits accuracy, specificity, sensitivity and MCC score of 0.824, 0.911, 0.831 and 0.663 respectively. The proposed model BBB-PEP-Prediction shown superlative performance as compared to previous benchmark studies. The proposed system helps in future research and research community for in-silico identification of BBB peptides.
Collapse
Affiliation(s)
- Ansar Naseem
- Department of Artificial Intelligence, School of Systems and Technology, University of Management and Technology, Lahore, Pakistan
| | - Fahad Alturise
- Department of Computer, College of Science and Arts in Ar Rass, Qassim University, Ar Rass, Saudi Arabia.
| | - Tamim Alkhalifah
- Department of Computer, College of Science and Arts in Ar Rass, Qassim University, Ar Rass, Saudi Arabia
| | - Yaser Daanial Khan
- Department of Computer Science, School of Systems and Technology, University of Management and Technology, Lahore, Pakistan
| |
Collapse
|
7
|
Gromiha MM, Harini K. Comment on 'Thermodynamic database supports deciphering protein-nucleic acid interactions'. Trends Biotechnol 2023; 41:988-989. [PMID: 37117054 DOI: 10.1016/j.tibtech.2023.03.014] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 03/22/2023] [Accepted: 03/22/2023] [Indexed: 04/30/2023]
Abstract
Mei and colleagues reported a thermodynamic database, PNATDB for protein-nucleic acid interactions, which contains 12 635 experimentally determined thermodynamic parameters. They claimed that extracting data from existing databases is difficult. ProNAB, which has more than 20 000 experimental data points for binding affinities of protein-nucleic acid complexes and other information, was not discussed.
Collapse
Affiliation(s)
- M Michael Gromiha
- Department of Biotechnology, Bhupat and Jyoti Mehta School of Biosciences, Indian Institute of Technology Madras, Chennai 600036, India.
| | - Kannan Harini
- Department of Biotechnology, Bhupat and Jyoti Mehta School of Biosciences, Indian Institute of Technology Madras, Chennai 600036, India
| |
Collapse
|
8
|
Attique M, Alkhalifah T, Alturise F, Khan YD. DeepBCE: Evaluation of deep learning models for identification of immunogenic B-cell epitopes. Comput Biol Chem 2023; 104:107874. [PMID: 37126975 DOI: 10.1016/j.compbiolchem.2023.107874] [Citation(s) in RCA: 2] [Impact Index Per Article: 2.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/16/2023] [Revised: 04/17/2023] [Accepted: 04/20/2023] [Indexed: 05/03/2023]
Abstract
B-Cell epitopes (BCEs) can identify and bind with receptor proteins (antigens) to initiate an immune response against pathogens. Understanding antigen-antibody binding interactions has many applications in biotechnology and biomedicine, including designing antibodies, therapeutics, and vaccines. Lab-based experimental identification of these proteins is time-consuming and challenging. Computational techniques have been proposed to discover BCEs, but most lack of significant accomplishments. This work uses classical and deep learning models (DLMs) with sequence-based features to predict immunity stimulator BCEs from proteomics sequences. The proposed convolutional neural network-based model outperforms other models with an accuracy (ACC) of 0.878, an F-measure of 0.871, and an area under the receiver operating characteristic curve (AUC) of 0.945. The proposed strategy achieves 58.7% better results on average than other state-of-the-art approaches based on the Mathews Correlation Coefficient (MCC) results. The established model is accessible through a web application located at http://deeplbcepred.pythonanywhere.com.
Collapse
Affiliation(s)
- Muhammad Attique
- Department of Computer Science, University of Management and Technology, Lahore 54000, Pakistan; Department of Information Technology, University of Gujrat, Gujrat 50700, Pakistan
| | - Tamim Alkhalifah
- Department of Computer, College of Science and Arts in Ar Rass Qassim University, Ar Rass, Qassim, Saudi Arabia.
| | - Fahad Alturise
- Department of Computer, College of Science and Arts in Ar Rass Qassim University, Ar Rass, Qassim, Saudi Arabia
| | - Yaser Daanial Khan
- Department of Computer Science, University of Management and Technology, Lahore 54000, Pakistan
| |
Collapse
|
9
|
Suleman MT, Alturise F, Alkhalifah T, Khan YD. iDHU-Ensem: Identification of dihydrouridine sites through ensemble learning models. Digit Health 2023; 9:20552076231165963. [PMID: 37009307 PMCID: PMC10064468 DOI: 10.1177/20552076231165963] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/29/2022] [Accepted: 03/09/2023] [Indexed: 04/04/2023] Open
Abstract
Background Dihydrouridine (D) is one of the most significant uridine modifications that have a prominent occurrence in eukaryotes. The folding and conformational flexibility of transfer RNA (tRNA) can be attained through this modification. Objective The modification also triggers lung cancer in humans. The identification of D sites was carried out through conventional laboratory methods; however, those were costly and time-consuming. The readiness of RNA sequences helps in the identification of D sites through computationally intelligent models. However, the most challenging part is turning these biological sequences into distinct vectors. Methods The current research proposed novel feature extraction mechanisms and the identification of D sites in tRNA sequences using ensemble models. The ensemble models were then subjected to evaluation using k-fold cross-validation and independent testing. Results The results revealed that the stacking ensemble model outperformed all the ensemble models by revealing 0.98 accuracy, 0.98 specificity, 0.97 sensitivity, and 0.92 Matthews Correlation Coefficient. The proposed model, iDHU-Ensem, was also compared with pre-existing predictors using an independent test. The accuracy scores have shown that the proposed model in this research study performed better than the available predictors. Conclusion The current research contributed towards the enhancement of D site identification capabilities through computationally intelligent methods. A web-based server, iDHU-Ensem, was also made available for the researchers at https://taseersuleman-idhu-ensem-idhu-ensem.streamlit.app/.
Collapse
Affiliation(s)
- Muhammad Taseer Suleman
- Department of Computer Science, School of systems and technology, University of Management and Technology, Lahore, Pakistan
| | - Fahad Alturise
- Department of Computer, College of Science and Arts in Ar Rass, Qassim University, Ar Rass, Qassim, Saudi Arabia
- Fahad Alturise, Department of Computer, College of Science and Arts in Ar Rass, Qassim University, Ar Rass, Qassim, Saudi Arabia.
| | - Tamim Alkhalifah
- Department of Computer, College of Science and Arts in Ar Rass, Qassim University, Ar Rass, Qassim, Saudi Arabia
| | - Yaser Daanial Khan
- Department of Computer Science, School of systems and technology, University of Management and Technology, Lahore, Pakistan
| |
Collapse
|
10
|
Akmal MA, Hassan MA, Muhammad S, Khurshid KS, Mohamed A. An analytical study on the identification of N-linked glycosylation sites using machine learning model. PeerJ Comput Sci 2022; 8:e1069. [PMID: 36262138 PMCID: PMC9575850 DOI: 10.7717/peerj-cs.1069] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 04/19/2022] [Accepted: 07/25/2022] [Indexed: 06/16/2023]
Abstract
N-linked is the most common type of glycosylation which plays a significant role in identifying various diseases such as type I diabetes and cancer and helps in drug development. Most of the proteins cannot perform their biological and psychological functionalities without undergoing such modification. Therefore, it is essential to identify such sites by computational techniques because of experimental limitations. This study aims to analyze and synthesize the progress to discover N-linked places using machine learning methods. It also explores the performance of currently available tools to predict such sites. Almost seventy research articles published in recognized journals of the N-linked glycosylation field have shortlisted after the rigorous filtering process. The findings of the studies have been reported based on multiple aspects: publication channel, feature set construction method, training algorithm, and performance evaluation. Moreover, a literature survey has developed a taxonomy of N-linked sequence identification. Our study focuses on the performance evaluation criteria, and the importance of N-linked glycosylation motivates us to discover resources that use computational methods instead of the experimental method due to its limitations.
Collapse
Affiliation(s)
- Muhammad Aizaz Akmal
- Department of Computer Science, University of Engineering and Technology, KSK, Lahore, Punjab, Pakistan
| | - Muhammad Awais Hassan
- Department of Computer Science, University of Engineering and Technology, Lahore, Punjab, Pakistan
| | - Shoaib Muhammad
- Department of Computer Science, University of Engineering and Technology, Lahore, Punjab, Pakistan
| | - Khaldoon S. Khurshid
- Department of Computer Science, University of Engineering and Technology, Lahore, Punjab, Pakistan
| | | |
Collapse
|
11
|
Comparative Analysis on Alignment-Based and Pretrained Feature Representations for the Identification of DNA-Binding Proteins. COMPUTATIONAL AND MATHEMATICAL METHODS IN MEDICINE 2022; 2022:5847242. [PMID: 35799660 PMCID: PMC9256349 DOI: 10.1155/2022/5847242] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Subscribe] [Scholar Register] [Received: 05/10/2022] [Accepted: 06/07/2022] [Indexed: 11/17/2022]
Abstract
The interaction between DNA and protein is vital for the development of a living body. Previous numerous studies on in silico identification of DNA-binding proteins (DBPs) usually include features extracted from the alignment-based (pseudo) position-specific scoring matrix (PSSM), leading to limited application due to its time-consuming generation. Few researchers have paid attention to the application of pretrained language models at the scale of evolution to the identification of DBPs. To this end, we present comprehensive insights into a comparison study on alignment-based PSSM and pretrained evolutionary scale modeling (ESM) representations in the field of DBP classification. The comparison is conducted by extracting information from PSSM and ESM representations using four unified averaging operations and by performing various feature selection (FS) methods. Experimental results demonstrate that the pretrained ESM representation outperforms the PSSM-derived features in a fair comparison perspective. The pretrained feature presentation deserves wide application to the area of in silico DBP identification as well as other function annotation issues. Finally, it is also confirmed that an ensemble scheme by aggregating various trained FS models can significantly improve the classification performance of DBPs.
Collapse
|