Abstract:Exploration of the intricate connections between long noncoding RNA (lncRNA) and diseases, referred to as lncRNA-disease associations (LDAs), plays a pivotal and indispensable role in unraveling the underlying molecular mechanisms of diseases and devising practical treatment approaches. It is imperative to employ computational methods for predicting lncRNA-disease associations to circumvent the need for superfluous experimental endeavors. Graph-based learning models have gained substantial popularity in predicting these associations, primarily because of their capacity to leverage node attributes and relationships within the network. Nevertheless, there remains much room for enhancing the performance of these techniques by incorporating and harmonizing the node attributes more effectively. In this context, we introduce a novel model, i.e., Adaptive Message Passing and Feature Fusion (AMPFLDAP), for forecasting lncRNA-disease associations within a heterogeneous network. Firstly, we constructed a heterogeneous network involving lncRNA, microRNA (miRNA), and diseases based on established associations and employing Gaussian interaction profile kernel similarity as a measure. Then, an adaptive topological message passing mechanism is suggested to address the information aggregation for heterogeneous networks. The topological features of nodes in the heterogeneous network were extracted based on the adaptive topological message passing mechanism. Moreover, an attention mechanism is applied to integrate both topological and semantic information to achieve the multimodal features of biomolecules, which are further used to predict potential LDAs. The experimental results demonstrated that the performance of the proposed AMPFLDAP is superior to seven state-of-the-art methods. Furthermore, to validate its efficacy in practical scenarios, we conducted detailed case studies involving three distinct diseases, which conclusively demonstrated AMPFLDAP's effectiveness in the prediction of LDAs.

Drug-disease association prediction using semantic graph and function similarity representation learning over heterogeneous information networks

RLFDDA: a meta-path based graph representation learning model for drug–disease association prediction

HINGRL: predicting drug–disease associations with graph representation learning on heterogeneous information networks

An effective drug-disease associations prediction model based on graphic representation learning over multi-biomolecular network

Learning from low-rank multimodal representations for predicting disease-drug associations

Drug-Disease Association Prediction Using Heterogeneous Networks for Computational Drug Repositioning

Similarity measures-based graph co-contrastive learning for drug–disease association prediction

REDDA: Integrating multiple biological relations to heterogeneous graph neural network for drug-disease association prediction

Regulation-aware graph learning for drug repositioning over heterogeneous biological network

Predicting drug-disease associations with heterogeneous network embedding.

Predicting Potential Drug-Disease Associations Based on Hypergraph Learning with Subgraph Matching

SAEROF: an ensemble approach for large-scale drug-disease association prediction by incorporating rotation forest and sparse autoencoder deep neural network

Hierarchical Negative Sampling Based Graph Contrastive Learning Approach for Drug-Disease Association Prediction

HNet-DNN: Inferring New Drug-Disease Associations with Deep Neural Network Based on Heterogeneous Network Features.

MGRL: Predicting Drug-Disease Associations Based on Multi-Graph Representation Learning

A geometric deep learning framework for drug repositioning over heterogeneous information networks

Gndd: A Graph Neural Network-Based Method For Drug-Disease Association Prediction

AMPFLDAP: Adaptive Message Passing and Feature Fusion on Heterogeneous Network for LncRNA-Disease Associations Prediction

Prediction of drug–disease associations based on reinforcement symmetric metric learning and graph convolution network

Prediction of Drug-Disease Associations Based on Multi-Kernel Deep Learning Method in Heterogeneous Graph Embedding

Drug-disease association prediction with literature based multi-feature fusion