SCPLPA: An miRNA-disease association prediction model based on spatial consistency projection and label propagation...

Area of Science:

• Biomedical informatics
• Computational biology
• Genomics

Background:

• Identifying microRNA (miRNA) and disease associations is crucial for disease prevention, diagnosis, and treatment.
• Existing computational methods often suffer from low prediction accuracy and weak generalization.
• There is a need for improved computational tools to predict human miRNA-disease associations.

Purpose of the Study:

• To propose a novel computational method, SCPLPA, for predicting miRNA-disease associations.
• To address the limitations of existing methods in terms of accuracy and generalization.
• To provide a reliable auxiliary tool for biomedical research.

Main Methods:

• Constructed heterogeneous disease and miRNA similarity networks using semantic, functional, and Gaussian interaction spectrum kernel similarities.
• Integrated label propagation algorithms on these heterogeneous networks to estimate miRNA-disease association scores.
• Employed a spatial consistency projection algorithm for feature extraction to predict unverified miRNA-disease associations.

Main Results:

• SCPLPA demonstrated superior predictive performance compared to four classical methods (MDHGI, NSEMDA, RFMDA, SNMFMDA) across multiple evaluation metrics.
• Case studies successfully identified miRNAs associated with colon neoplasms and kidney neoplasms.
• The algorithm proved effective in predicting unverified miRNA-disease associations.

Conclusions:

• SCPLPA is an easy-to-implement algorithm that effectively predicts miRNA-disease associations.
• The method offers improved accuracy and generalization over existing computational approaches.
• SCPLPA serves as a valuable auxiliary tool for advancing biomedical research in miRNA-disease associations.