Explainable deep neural networks for predicting sample phenotypes from single-cell transcriptomics DOI Creative Commons
Jordi Martorell‐Marugán, Raúl López-Domínguez, Juan Antonio Villatoro-García

et al.

Briefings in Bioinformatics, Journal Year: 2024, Volume and Issue: 26(1)

Published: Nov. 22, 2024

Abstract Recent advances in single-cell RNA-Sequencing (scRNA-Seq) technologies have revolutionized our ability to gather molecular insights into different phenotypes at the level of individual cells. The analysis resulting data poses significant challenges, and proper statistical methods are required analyze extract information from scRNA-Seq datasets. Sample classification based on gene expression has proven effective valuable for precision medicine applications. However, standard schemas often not suitable due their unique characteristics, new algorithms effectively classify samples level. Furthermore, existing this purpose limitations usability. Those reasons motivated us develop singleDeep, an end-to-end pipeline that streamlines training deep neural networks, enabling robust prediction characterization sample phenotypes. We used singleDeep make predictions datasets conditions, including systemic lupus erythematosus, Alzheimer’s disease coronavirus 2019. Our results demonstrate strong diagnostic performance, validated both internally externally. Moreover, outperformed traditional machine learning alternative approaches. In addition accuracy, provides cell types importance estimation phenotypic characterization. This functionality provided additional use cases. For instance, we corroborated some interferon signature genes consistently relevant autoimmunity across all immune lupus. On other hand, discovered linked dementia roles specific brain populations, such as APOE astrocytes.

Language: Английский

DTI-MHAPR: optimized drug-target interaction prediction via PCA-enhanced features and heterogeneous graph attention networks DOI Creative Commons
Guang Yang, Yinbo Liu,

SiJian Wen

et al.

BMC Bioinformatics, Journal Year: 2025, Volume and Issue: 26(1)

Published: Jan. 13, 2025

Drug-target interactions (DTIs) are pivotal in drug discovery and development, their accurate identification can significantly expedite the process. Numerous DTI prediction methods have emerged, yet many fail to fully harness feature information of drugs targets or address issue redundancy. We aim refine accuracy by eliminating redundant features capitalizing on node topological structure enhance extraction. To achieve this, we introduce a PCA-augmented multi-layer heterogeneous graph-based network that concentrates key throughout encoding-decoding phase. Our approach initiates with construction graph from various similarity metrics, which is then encoded via neural network. concatenate integrate resultant representation vectors merge multi-level information. Subsequently, principal component analysis applied distill most informative features, random forest algorithm employed for final decoding integrated data. method outperforms six baseline models terms accuracy, as demonstrated extensive experimentation. Comprehensive ablation studies, visualization results, in-depth case analyses further validate our framework's efficacy interpretability, providing novel tool integrates multimodal features.

Language: Английский

Citations

1
RNA structure prediction using deep learning — A comprehensive review DOI Creative Commons
Mayank Chaturvedi, Mahmood A. Rashid, Kuldip K. Paliwal

et al.

Computers in Biology and Medicine, Journal Year: 2025, Volume and Issue: 188, P. 109845 - 109845

Published: Feb. 20, 2025

In computational biology, accurate RNA structure prediction offers several benefits, including facilitating a better understanding of functions and RNA-based drug design. Implementing deep learning techniques for has led tremendous progress in this field, resulting significant improvements accuracy. This comprehensive review aims to provide an overview the diverse strategies employed predicting secondary structures, emphasizing methods. The article categorizes discussion into three main dimensions: feature extraction methods, existing state-of-the-art model architectures, approaches. We present comparative analysis various models highlighting their strengths weaknesses. Finally, we identify gaps literature, discuss current challenges, suggest future approaches enhance performance applicability tasks. provides deeper insight subject paves way further dynamic intersection life sciences artificial intelligence.

Language: Английский

Citations

1
Editorial: Machine Learning in Bio-cheminformatics DOI Open Access
Kenneth M. Merz, Guo‐Wei Wei, Feng Zhu

et al.

Journal of Chemical Information and Modeling, Journal Year: 2024, Volume and Issue: 64(7), P. 2125 - 2128

Published: April 8, 2024

ADVERTISEMENT RETURN TO ISSUEEditorialNEXTEditorial: Machine Learning in Bio-cheminformaticsKenneth M. Merz*Kenneth MerzDepartment of Chemistry, Michigan State University, Lansing 48824, Michigan, United States*(Email: [email protected]).More by Kenneth Merzhttps://orcid.org/0000-0001-9139-5893, Guo-Wei Wei*Guo-Wei WeiDepartment Mathematics, Weihttps://orcid.org/0000-0002-5781-2937, and Feng Zhu*Feng ZhuCollege Pharmaceutical Sciences, Zhejiang Hangzhou 310058, Zhejiang, China*(Email: Zhuhttps://orcid.org/0000-0001-8069-0053Cite this: J. Chem. Inf. Model. 2024, 64, 7, 2125–2128Publication Date (Web):April 8, 2024Publication History Received13 March 2024Published online8 April inissue 8 2024https://pubs.acs.org/doi/10.1021/acs.jcim.4c00444https://doi.org/10.1021/acs.jcim.4c00444editorialACS PublicationsCopyright © 2024 American Chemical Society. This publication is available under these Terms Use. Request reuse permissions free to access through this site. Learn MoreArticle Views1288Altmetric-Citations-LEARN ABOUT THESE METRICSArticle Views are the COUNTER-compliant sum full text article downloads since November 2008 (both PDF HTML) across all institutions individuals. These metrics regularly updated reflect usage leading up last few days.Citations number other articles citing article, calculated Crossref daily. Find more information about citation counts.The Altmetric Attention Score a quantitative measure attention that research has received online. Clicking on donut icon will load page at altmetric.com with additional details score social media presence for given article. how calculated. Share Add toView InAdd Full Text ReferenceAdd Description ExportRISCitationCitation abstractCitation referencesMore Options onFacebookTwitterWechatLinked InRedditEmail (1 MB) Get e-AlertscloseSUBJECTS:Bioinformatics computational biology,Chemoinformatics,Genetics,Peptides proteins,Protein structure e-Alerts

Language: Английский

Citations

5
Deep learning for predicting synergistic drug combinations: State‐of‐the‐arts and future directions DOI Creative Commons
Yu Wang, Junjie Wang, Yun Liu

et al.

Clinical and Translational Discovery, Journal Year: 2024, Volume and Issue: 4(3)

Published: June 1, 2024

Abstract Combination therapy has emerged as an efficacy strategy for treating complex diseases. Its potential to overcome drug resistance and minimize toxicity makes it highly desirable. However, the vast number of pairs presents a significant challenge, rendering exhaustive clinical testing impractical. In recent years, deep learning‐based methods have promising tools predicting synergistic combinations. This review aims provide comprehensive overview applying diverse deep‐learning architectures combination prediction. commences by elucidating quantitative measures employed assess synergy. Subsequently, we delve into various currently Finally, concludes outlining key challenges facing learning approaches proposes future research.

Language: Английский

Citations

4
Variable multi-scale attention fusion network and adaptive correcting gradient optimization for multi-task learning DOI

Naihua Ji,

Yongqiang Sun,

Fanyun Meng

et al.

Pattern Recognition, Journal Year: 2025, Volume and Issue: unknown, P. 111423 - 111423

Published: Feb. 1, 2025

Language: Английский

Citations

0
MGTNSyn: Molecular structure-aware graph transformer network with relational attention for drug synergy prediction DOI
Yunjiong Liu, Peiliang Zhang, Dongyang Li

et al.

Expert Systems with Applications, Journal Year: 2025, Volume and Issue: unknown, P. 127699 - 127699

Published: April 1, 2025

Language: Английский

Citations

0
Characterizing Clinical Toxicity in Cancer Combination Therapies DOI Creative Commons
Alexandra M. Wong, Lorin Crawford

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: April 19, 2025

A bstract Predicting synergistic cancer drug combinations through computational methods offers a scalable approach to creating therapies that are more effective and less toxic. However, most algorithms focus solely on synergy without considering toxicity when selecting optimal combinations. In the absence of combinatorial assays, few models use penalties balance high with lower toxicity. these have not been explicitly validated against known drug-drug interactions. this study, we examine whether scores metrics correlate adverse While some show trends levels, our results reveal significant limitations in using them as penalties. These findings highlight challenges incorporating into prediction frameworks suggest advancing field requires comprehensive combination data.

Language: Английский

Citations

0
Drug–target interaction prediction based on improved heterogeneous graph representation learning and feature projection classification DOI
Donghua Yu, Huawen Liu, Shuang Yao

et al.

Expert Systems with Applications, Journal Year: 2024, Volume and Issue: 252, P. 124289 - 124289

Published: May 27, 2024

Language: Английский

Citations

3
GS-DTA: integrating graph and sequence models for predicting drug-target binding affinity DOI Creative Commons
Junwei Luo, Zheng Zhu, Zhiyu Xu

et al.

BMC Genomics, Journal Year: 2025, Volume and Issue: 26(1)

Published: Feb. 4, 2025

Drug-target binding affinity (DTA) prediction is vital in drug discovery and repositioning, more researchers are beginning to focus on this. Many effective methods have been proposed. However, some current certain shortcomings focusing important nodes molecular graphs dealing with complex structural molecules. In particular, when considering substructures molecules, they may not be able fully explore the potential relationships between different parts. addition, protein structures, ignore connections amino acid fragments that far apart sequence but work synergistically function. this paper, we propose a new method, called GS-DTA, for predicting DTA based graph models. GS-DTA takes simplified input line system (SMILES) of as input. First, each modeled graph, which vertex an atom edge represents interaction atoms. Then GATv2-GCN three-layer GCN networks used extract features drug. enhances model's ability by assigning dynamic attention scores, improves learning structure's intricate patterns. Besides, The can captures hierarchical through deeper propagation feature transformation. Meanwhile, protein, framework combining CNN, Bi-LSTM, Transformer contextual information sequences, combination help understand comprehensive detailed protein. Finally, obtained vectors combined predict connected layer. source code downloaded from https://github.com/zhuziguang/GS-DTA . results show achieves good performance terms MSE, CI, r2m Davis KIBA datasets, improving accuracy prediction.

Language: Английский

Citations

0
DeepGRNCS: deep learning-based framework for jointly inferring gene regulatory networks across cell subpopulations DOI Creative Commons

Yahui Lei,

Xiaotai Huang, Xingli Guo

et al.

Briefings in Bioinformatics, Journal Year: 2024, Volume and Issue: 25(4)

Published: May 23, 2024

Abstract Inferring gene regulatory networks (GRNs) allows us to obtain a deeper understanding of cellular function and disease pathogenesis. Recent advances in single-cell RNA sequencing (scRNA-seq) technology have improved the accuracy GRN inference. However, many methods for inferring individual GRNs from scRNA-seq data are limited because they overlook intercellular heterogeneity similarities between different cell subpopulations, which often present data. Here, we propose deep learning-based framework, DeepGRNCS, jointly across subpopulations. We follow commonly accepted hypothesis that expression target can be predicted based on transcription factors (TFs) due underlying relationships. initially processed by discretizing scattering using equal-width method. Then, trained learning models predict TFs. By individually removing each TF matrix, used pre-trained model predictions infer relationships TFs genes, thereby constructing GRN. Our method outperforms existing inference various simulated real datasets. Finally, applied DeepGRNCS non-small lung cancer identify key genes subpopulation analyzed their biological relevance. In conclusion, effectively predicts subpopulation-specific GRNs. The source code is available at https://github.com/Nastume777/DeepGRNCS.

Language: Английский

Citations

2