MultiSourcDSim: an integrated approach for exploring disease similarity

Area of Science:

• Computational biology
• Bioinformatics
• Medical informatics

Background:

• Understanding disease associations is key to uncovering common pathogenic mechanisms.
• Identifying disease similarity can inform cross-disease treatment strategies.
• Existing methods for disease similarity often lack accuracy due to reliance on limited data sources.

Purpose of the Study:

• To propose a novel method, MultiSourcDSim, for calculating disease similarity.
• To improve the accuracy of disease similarity prediction by integrating multiple data sources.

Main Methods:

• Integrated gene-disease associations, GO biological process-disease associations, and symptom-disease associations.
• Constructed three distinct disease similarity networks.
• Developed a node representation by integrating these networks for similarity calculation.

Main Results:

• The MultiSourcDSim approach demonstrated superior performance compared to three other popular methods.
• The generated disease similarity network revealed previously unknown latent relationships between diseases.

Conclusions:

• MultiSourcDSim provides an efficient and accurate method for predicting disease similarity.
• The method effectively integrates diverse data sources to uncover complex disease interdependencies.