Explainable Artificial Intelligence for Protein Function Prediction: A Perspective View

Article Biochemical Research Methods

Deep transformers and convolutional neural network in identifying DNA N6-methyladenine sites in cross-species genomes

Nguyen Quoc Khanh Le et al.

Summary: This study presents a novel model based on the transformer architecture and deep learning to accurately identify DNA 6mA sites from cross-species genomes. The model achieved excellent performance in independent testing, with an accuracy of 79.3% and a Matthews correlation coefficient (MCC) of 0.58. It outperformed baseline models and existing predictors, demonstrating the effectiveness of the proposed hybrid framework.

METHODS (2022)

添加到收藏夹

Article Biochemical Research Methods

Effector-GAN: prediction of fungal effector proteins based on pretrained deep representation learning methods and generative adversarial networks

Yansu Wang et al.

Summary: In this study, deep representation learning methods and generative adversarial networks were used for fungal effector protein prediction, resulting in improved accuracy compared to previous methods.

BIOINFORMATICS (2022)

添加到收藏夹

Article Biochemical Research Methods

A Novel Method for Predicting Essential Proteins by Integrating Multidimensional Biological Attribute Information and Topological Properties

Hanyu Lu et al.

Summary: This paper proposes a new method (EOP) for inferring potential essential proteins by combining the multidimensional biological attribute information of proteins with the topological properties of the protein-protein interaction network. The simulation results show that this method identifies more essential proteins compared to other methods, and has higher recognition rates under various conditions.

CURRENT BIOINFORMATICS (2022)

添加到收藏夹

Article Computer Science, Information Systems

Explainable Deep Learning Approach for Multilabel Classification of Antimicrobial Resistance With Missing Labels

Mukunthan Tharmakulasingam et al.

Summary: This paper presents an effective deep learning solution, ML-ConvNet, for predicting antimicrobial resistance (AMR) on datasets with missing labels. The proposed model outperforms existing methods and is equipped with an explainable artificial intelligence pipeline, making it suitable for healthcare settings.

IEEE ACCESS (2022)

添加到收藏夹

Article Biochemical Research Methods

Prediction of FMN Binding Sites in Electron Transport Chains Based on 2-D CNN and PSSM Profiles

Nguyen-Quoc-Khanh Le et al.

Summary: Flavin mono-nucleotides (FMNs) are crucial cofactors in cellular respiration, responsible for electron transport chain; understanding their functions can provide insights into human diseases and drug targets. A deep learning model was proposed to identify FMN interacting residues with high accuracy and sensitivity, outperforming previous studies and showing potential for broader applications in bioinformatics and computational biology.

IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS (2021)

添加到收藏夹

Article Biochemical Research Methods

Prediction Model of Thermophilic Protein Based on Stacking Method

Xian-Fang Wang et al.

Summary: A thermophilic protein prediction model based on the Stacking method was proposed in this study, achieving an accuracy of up to 93.75% when verified by the Jackknife method. The overall performance was better than most reported methods, showing strong robustness and significantly improving the prediction performance of thermophilic proteins.

CURRENT BIOINFORMATICS (2021)

添加到收藏夹

Article Multidisciplinary Sciences

Highly accurate protein structure prediction with AlphaFold

John Jumper et al.

Summary: Proteins are essential for life, and accurate prediction of their structures is a crucial research problem. Current experimental methods are time-consuming, highlighting the need for accurate computational approaches to address the gap in structural coverage. Despite recent progress, existing methods fall short of atomic accuracy in protein structure prediction.

NATURE (2021)

添加到收藏夹

Article Biotechnology & Applied Microbiology

An explainable artificial intelligence approach for decoding the enhancer histone modifications code and identification of novel enhancers in Drosophila

Jareth C. Wolfe et al.

Summary: By using machine learning and explainable artificial intelligence model, the location of known enhancers in Drosophila can be predicted accurately, providing insight into the underlying histone modifications code. A large set of putative enhancers with similar epigenetic signature as known enhancers were identified, showing intermediate enrichment of mediator and cohesin complexes. 10-15% of predicted enhancers display characteristics similar to super enhancers observed in other species.

GENOME BIOLOGY (2021)

添加到收藏夹

Article Computer Science, Artificial Intelligence