SFPEL-LPI: Sequence-based feature projection ensemble learning for predicting LncRNA-protein interactions.

Area of Science:

• Computational Biology
• Bioinformatics
• Molecular Biology

Background:

• Long non-coding RNA (lncRNA)-protein interactions are crucial for gene regulation, including post-transcriptional modification, splicing, and translation.
• Understanding these interactions is key to deciphering lncRNA functions.
• Current prediction methods often require extensive features and struggle with novel or uncharacterized lncRNAs and proteins.

Purpose of the Study:

• To develop a robust computational method for predicting lncRNA-protein interactions.
• To address the limitations of existing methods in handling new or unannotated lncRNAs and proteins.
• To provide a user-friendly tool for exploring lncRNA-protein associations.

Main Methods:

• Proposed SFPEL-LPI, a sequence-based feature projection ensemble learning method.
• Extracted sequence-based features from both lncRNAs and proteins.
• Calculated lncRNA-lncRNA and protein-protein similarities using sequences and known interactions, integrated within an ensemble learning framework.

Main Results:

• SFPEL-LPI demonstrated high accuracy in predicting lncRNA-protein associations.
• The method outperformed existing state-of-the-art approaches in computational experiments.
• Successfully predicted novel lncRNA-protein interactions, validated by literature.

Conclusions:

• SFPEL-LPI is an effective and versatile tool for predicting lncRNA-protein interactions, including for novel molecules.
• The developed web server (http://www.bioinfotech.cn/SFPEL-LPI/) provides accessible prediction capabilities.
• This work advances the understanding of lncRNA functions through accurate interaction prediction.