TCSRWRLD

TCSRWRLD predicts potential associations between long non-coding RNAs (lncRNAs) and diseases by applying an improved random walk with restart on a heterogeneous lncRNA–disease similarity network to prioritize candidate links.

Key Features:

• Heterogeneous network construction: constructs a heterogeneous network representing lncRNA and disease nodes by integrating similarities of both lncRNAs and diseases.
• Target Convergence Set (TCS): defines for each lncRNA or disease node a TCS composed of the top 100 nodes with minimum average network distances to nodes already known to have associations.
• Improved random walk with restart: applies an improved random walk with restart that leverages the heterogeneous network and the TCS and halts when stable state probabilities are reached at TCS nodes to accelerate convergence.

Scientific Applications:

• Prediction of lncRNA–disease associations: demonstrated AUC = 0.8712 in Leave-One-Out Cross Validation (LOOCV) for predicting lncRNA-disease associations.
• Case studies: validated predictive capability in case studies of lung cancer and leukemia.

Methodology:

Construct a heterogeneous lncRNA–disease network by integrating lncRNA and disease similarities; for each node compute a TCS of the top 100 nodes with minimum average network distances to known associated nodes; run an improved random walk with restart on the network using the TCS and halt when stable state probabilities are achieved at TCS nodes.