LPI-SKF

LPI-SKF predicts interactions between long non-coding RNAs (lncRNAs) and proteins by integrating multiple similarity measures using Similarity Kernel Fusion (SKF) and Laplacian Regularized Least Squares (LapRLS).

Key Features:

• Integration of Multiple Similarities: Incorporates diverse similarity metrics for both lncRNAs and proteins to capture a comprehensive range of interaction features.
• Advanced Algorithms: Applies Similarity Kernel Fusion (SKF) to fuse kernel matrices and Laplacian Regularized Least Squares (LapRLS) to regularize the prediction process.
• High Predictive Accuracy: Validated using 5-fold cross-validation and reported an AUROC of 0.909.
• Verification of Predictions: Nineteen of the top twenty predicted lncRNA–protein interactions were corroborated by existing data.

Scientific Applications:

• Functional mechanism inference: Predicts lncRNA–protein interactions to support inference of lncRNA roles in transcription, splicing, and translation regulation.
• Novel interaction discovery: Enables prediction of interactions involving novel proteins or novel lncRNAs to prioritize candidates for experimental study.

Methodology:

Integrates diverse similarity measures for lncRNAs and proteins; fuses multiple kernel matrices using Similarity Kernel Fusion (SKF); applies Laplacian Regularized Least Squares (LapRLS) for regularized prediction; validated by 5-fold cross-validation.