Data resources and computational methods for lncRNA-disease association prediction.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Long non-coding RNAs (lncRNAs) play diverse roles in genome regulation, influencing disease pathogenesis.
• Understanding lncRNA-disease associations (LDAs) is crucial for deciphering disease mechanisms.
• Computational models significantly aid systematic biology and experimental research in this field.

Purpose of the Study:

• To review representative diseases linked to lncRNAs, including cancers, cardiovascular, and neurological diseases.
• To introduce publicly available resources for lncRNA and disease data.
• To systematically categorize and discuss computational methods for LDA prediction.

Main Methods:

• Categorization of 64 computational LDA prediction methods into five groups: machine learning, network propagation, matrix factorization/completion, deep learning, and graph neural networks.
• Discussion of common evaluation methods and metrics used in LDA prediction.
• Identification of current challenges and future trends in the field.

Main Results:

• A comprehensive overview of 64 computational methods for LDA prediction, classified into distinct categories.
• Introduction to relevant public resources for lncRNA-disease research.
• Analysis of evaluation strategies and metrics for assessing prediction accuracy.

Conclusions:

• LDA prediction is a rapidly advancing field with significant implications for understanding disease pathogenesis.
• The review provides a structured overview of existing computational approaches and resources.
• Future research should focus on addressing current challenges and leveraging emerging trends in LDA prediction.