LncRNA-disease association identification using graph auto-encoder and learning to rank.

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Identifying long non-coding RNA (lncRNA)-disease associations is vital for disease treatment, diagnosis, and prevention.
• Wet laboratory experiments are costly and labor-intensive, necessitating efficient computational prediction methods.
• Existing methods combining classification and Learning to Rank (LTR) can be affected by incorrect classification results.

Purpose of the Study:

• To develop an efficient computational method for predicting potential lncRNA-disease associations.
• To improve the accuracy of lncRNA-disease association prediction by leveraging biological knowledge graphs and a ranking framework.
• To overcome limitations of previous methods by integrating graph-based feature extraction with LTR.

Main Methods:

• Constructing homogeneous and heterogeneous graphs by integrating multi-source biological information.
• Utilizing a graph auto-encoder and attention mechanism to extract embedded features from constructed graphs.
• Integrating extracted features into the LTR framework using feature crossing statistical strategies to predict disease association priorities for lncRNAs.

Main Results:

• The proposed GraLTR-LDA predictor demonstrated superior performance compared to state-of-the-art methods.
• GraLTR-LDA effectively identified potential long non-coding RNA-disease associations.
• Experimental results validated the efficacy of the graph-based feature extraction and LTR integration.

Conclusions:

• GraLTR-LDA offers an effective computational approach for predicting lncRNA-disease associations.
• The method enhances the accuracy and efficiency of discovering novel lncRNA-disease relationships.
• Availability of datasets and source codes facilitates further research and application in the field.