Multivariate Information Fusion With Fast Kernel Learning to Kernel Ridge Regression in Predicting LncRNA-Protein...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Long non-coding RNAs (lncRNAs) are key regulators of gene expression.
• Predicting lncRNA-protein interactions (LPI) is crucial but challenging with traditional experimental methods.
• Existing computational LPI prediction methods require improvement in accuracy and speed.

Purpose of the Study:

• To develop a novel and efficient computational method for predicting lncRNA-protein interactions (LPI).
• To enhance the accuracy and velocity of LPI prediction using machine learning and integrated similarity measures.

Main Methods:

• Proposed LPI-FKLKRR method utilizing Kernel Ridge Regression (KRR) and Fast Kernel Learning (FastKL).
• Employed four distinct similarity measures for lncRNA and protein spaces.
• Integrated Gene Ontology (GO) information for proteins to improve data quality.

Main Results:

• LPI-FKLKRR achieved a superior Area Under Precision Recall Curve (AUPR) of 0.6950 on a benchmark dataset.
• Outperformed existing LPI prediction methods, including LPLNP, RWR, CF, LPIHN, and LPBNI.
• Demonstrated strong performance in a case study on a novel dataset.

Conclusions:

• LPI-FKLKRR offers a significant advancement in computational LPI prediction.
• The method provides a valuable tool for researchers studying gene expression regulation.
• Highlights the potential of integrating diverse similarity measures and machine learning for biological predictions.