CCLDA: prediction of lncRNA-disease associations based on Convolutional Block Attention Module and Capsule Network

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Long non-coding RNAs (lncRNAs) play critical roles in various disease processes.
• Identifying lncRNA-disease associations (LDAs) is vital for disease diagnosis and treatment.
• Computational methods offer efficient and cost-effective alternatives to experimental LDA identification.

Purpose of the Study:

• To introduce CCLDA, a novel deep learning model for predicting lncRNA-disease associations.
• To improve the accuracy and efficiency of LDA prediction using advanced computational techniques.

Main Methods:

• Constructed similarity matrices (functional, Gaussian, sequence for lncRNAs; semantic, Gaussian for diseases) and fused them.
• Employed a multilayer autoencoder for feature extraction from lncRNA-disease pairs.
• Utilized a capsule network integrated with a Convolutional Attention Module (CBAM) for prediction.

Main Results:

• CCLDA demonstrated superior performance compared to existing methods on two benchmark datasets.
• Ablation experiments validated the contribution of individual components within the CCLDA model.
• Case studies highlighted CCLDA's potential for discovering novel LDAs.

Conclusions:

• CCLDA is an effective deep learning approach for predicting lncRNA-disease associations.
• The model shows significant promise for advancing research in lncRNA-disease prediction and related fields.
• CCLDA offers a valuable tool for enhancing disease diagnosis and therapeutic strategies.