MCA-Net: Multi-Feature Coding and Attention Convolutional Neural Network for Predicting lncRNA-Disease Association.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Traditional methods for predicting long non-coding RNA (lncRNA)-disease associations are time-consuming and subjective.
• The era of big data necessitates efficient and accurate computational approaches for identifying lncRNA-disease relationships.

Purpose of the Study:

• To develop a novel deep learning method for predicting lncRNA-disease associations.
• To improve the accuracy and efficiency of lncRNA-disease association prediction compared to existing methods.

Main Methods:

• Calculated six similarity features to extract lncRNA and disease information.
• Proposed a multi-feature coding method to construct feature vectors for lncRNA-disease association samples.
• Developed an attention convolutional neural network (MCA-Net) for prediction using 10-fold cross-validation.

Main Results:

• MCA-Net demonstrated superior performance over state-of-the-art methods on three public datasets (LncRNADisease, Lnc2Cancer, LncRNADisease2.0).
• Model parameter effects, comparisons with other deep learning models, and the necessity of the attention mechanism were evaluated.
• Case studies on breast and lung cancer confirmed MCA-Net's effectiveness and accuracy.

Conclusions:

• MCA-Net is an effective and accurate deep learning model for predicting lncRNA-disease associations.
• The proposed method offers a significant advancement in computational approaches for understanding lncRNA roles in diseases.