ncRNALocate-EL: a multi-label ncRNA subcellular locality prediction model based on ensemble learning.

Area of Science:

• Molecular Biology
• Bioinformatics
• Computational Biology

Background:

• Subcellular localization of non-coding RNAs (ncRNAs) is crucial for their biological functions.
• Current computational methods for predicting ncRNA localization have limitations, including outdated training data and inability to predict multiple localizations concurrently.

Purpose of the Study:

• To develop an improved computational model for predicting the subcellular localization of ncRNAs.
• To address the limitations of existing methods by enabling multi-label classification and utilizing up-to-date datasets.

Main Methods:

• Construction of three human ncRNA subcellular datasets (lncRNA, miRNA, snoRNA) using the RNALocate database.
• Development of a novel multi-label classification model, ncRNALocate-EL, based on ensemble learning.

Main Results:

• The proposed ncRNALocate-EL model demonstrates superior performance compared to previous methods.
• Achieved high average precision scores: 0.709 for lncRNA, 0.977 for miRNA, and 0.730 for snoRNA.
• A web server for ncRNALocate-EL is available for public use.

Conclusions:

• ncRNALocate-EL effectively predicts multi-label subcellular localizations for human ncRNAs.
• The model represents a significant advancement in computational ncRNA localization prediction.
• The accessible web server facilitates broader research application.