A novel algorithm based on bi-random walks to identify disease-related lncRNAs

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Long non-coding RNAs (lncRNAs) are implicated in various biological processes and diseases.
• Dysfunctional lncRNAs are linked to numerous diseases, necessitating accurate diagnostic tools.
• Existing computational models for lncRNA-disease association prediction have room for improvement.

Purpose of the Study:

• To develop a novel computational algorithm, BiWalkLDA, for predicting lncRNA-disease associations.
• To enhance the accuracy and specificity of lncRNA-disease association predictions.
• To provide a better understanding of the molecular mechanisms underlying diseases.

Main Methods:

• Developed BiWalkLDA algorithm utilizing bi-random walks.
• Constructed a lncRNA-disease network integrating interaction profiles and gene ontology information.
• Addressed the cold-start problem using neighbors' interaction profile information.

Main Results:

• BiWalkLDA demonstrated superior performance compared to existing algorithms (SIMCLDA, LDAP, LRLSLDA) in predicting lncRNA-disease associations.
• Validation on three real biological datasets showed improved accuracy and specificity.
• A case study on prostate cancer confirmed eight of the top-ten predicted lncRNAs in existing literature.

Conclusions:

• BiWalkLDA effectively predicts lncRNA-disease associations using bi-random walks.
• The algorithm offers enhanced accuracy and specificity over existing methods.
• BiWalkLDA provides a valuable tool for disease diagnosis and understanding molecular mechanisms.