CNNDLP: A Method Based on Convolutional Autoencoder and Convolutional Neural Network with Adjacent Edge Attention for...

Area of Science:

• Genomics and Bioinformatics
• Molecular Biology
• Computational Biology

Background:

• Aberrant expression of long non-coding RNAs (lncRNAs) is linked to disease pathogenesis.
• Accurate prediction of lncRNA-disease associations aids in understanding disease mechanisms.
• Existing shallow learning models struggle to integrate heterogeneous data and learn low-dimensional representations.

Purpose of the Study:

• To develop a deep learning model for predicting candidate disease-related lncRNAs.
• To integrate multi-source heterogeneous data including lncRNA-disease associations, interactions, and similarities.
• To improve the accuracy and depth of lncRNA-disease association prediction.

Main Methods:

• Proposed a novel method, CNNDLP, utilizing a convolutional neural network (CNN) with an attention mechanism and a convolutional autoencoder.
• Integrated heterogeneous data: lncRNA-disease associations, lncRNA-miRNA interactions, disease-miRNA interactions, and their respective similarities.
• Developed attention mechanisms at the adjacent edge level and calculated novel lncRNA and disease similarities based on network topology.

Main Results:

• CNNDLP demonstrated superior prediction performance over state-of-the-art methods in cross-validation experiments.
• The model effectively learned low-dimensional network representations and attention mechanisms for lncRNA-disease pairs.
• Case studies on stomach, breast, and prostate cancers validated CNNDLP's ability to discover potential disease-related lncRNAs.

Conclusions:

• CNNDLP offers a powerful deep learning framework for predicting lncRNA-disease associations.
• The method's ability to integrate diverse data and learn complex representations enhances understanding of molecular pathogenesis.
• CNNDLP shows promise for identifying novel biomarkers and therapeutic targets for various diseases.