Utilizing Dual-Channel Graph and Hypergraph Convolution Network to Discover Microbes Underlying Disease Traits DOI
Jing Chen,

Leyang Zhang,

Zhipan Liang

и другие.

Journal of Chemical Information and Modeling, Год журнала: 2025, Номер unknown

Опубликована: Май 15, 2025

Discovering microbes underlying disease traits opens up opportunities for the diagnosis and effective treatment of diseases. However, traditional methods are often based on biological experiments, which not only time-consuming but also costly, driving need computational frameworks that can accelerate discovery these associations. Motivated by challenges, we propose an innovative prediction algorithm named dual-channel graph Hypergraph Convolutional Network (DCGHCN) to discover traits. First, K-Nearest Neighbors (KNN) principle, constructed attribute graphs diseases, respectively. Next, Graph Networks (GCNs) used capture homogeneous level implicit representations from We output GCN layer as input construct a hypergraph convolutional evaluate impact confirmed diseases associations (MDAs) results. Perform scalar product calculation microbe features determine predicted score. The innovation DCGHCN lies in employing KNN handle missing values correlation matrix during preprocessing use structure combine advantages GCNs (HGCNs). 5-fold cross-validation (CV) performance DCGHCN. results showed model achieved AUC (Area Under ROC Curve), AUPR PR F1-score accuracy 0.9415, 0.7637, 0.7515, 0.9818. selected two case studies, large number published literature conclusions DCGHCN, thus proving is tool discovering

Язык: Английский

How Do Microbial Metabolites Interact with Their Protein Targets? DOI

Mario Astigarraga,

Andrés Sánchez-Ruiz,

Aminata Diop-Aw

и другие.

Journal of Chemical Information and Modeling, Год журнала: 2025, Номер 65(1), С. 201 - 213

Опубликована: Янв. 2, 2025

The design of drugs and nutraceutics that mimic microbial metabolites is an emerging drug modality in medicinal chemistry attempts to modulate the myriad interactions these molecules establish with host proteins. Understanding how interact their target proteins key perform a rational metabolite mimetic for therapeutic usage. In present work, we address this question by analyzing functional groups they display set more than 71K protein–metabolite from PDB. Significant differences group distributions, chemical features, co-occurrences are observed distinct subsets molecules. same true distributions interaction types. By correlating both data sets, able explain patterns terms patterns. These results will shed light on novel purposes.

Язык: Английский

Процитировано

1
Utilizing Dual-Channel Graph and Hypergraph Convolution Network to Discover Microbes Underlying Disease Traits DOI
Jing Chen,

Leyang Zhang,

Zhipan Liang

и другие.

Journal of Chemical Information and Modeling, Год журнала: 2025, Номер unknown

Опубликована: Май 15, 2025

Discovering microbes underlying disease traits opens up opportunities for the diagnosis and effective treatment of diseases. However, traditional methods are often based on biological experiments, which not only time-consuming but also costly, driving need computational frameworks that can accelerate discovery these associations. Motivated by challenges, we propose an innovative prediction algorithm named dual-channel graph Hypergraph Convolutional Network (DCGHCN) to discover traits. First, K-Nearest Neighbors (KNN) principle, constructed attribute graphs diseases, respectively. Next, Graph Networks (GCNs) used capture homogeneous level implicit representations from We output GCN layer as input construct a hypergraph convolutional evaluate impact confirmed diseases associations (MDAs) results. Perform scalar product calculation microbe features determine predicted score. The innovation DCGHCN lies in employing KNN handle missing values correlation matrix during preprocessing use structure combine advantages GCNs (HGCNs). 5-fold cross-validation (CV) performance DCGHCN. results showed model achieved AUC (Area Under ROC Curve), AUPR PR F1-score accuracy 0.9415, 0.7637, 0.7515, 0.9818. selected two case studies, large number published literature conclusions DCGHCN, thus proving is tool discovering

Язык: Английский

Процитировано

0