A systematic review of computational methods for predicting long noncoding RNAs.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Accurate identification of long noncoding RNAs (lncRNAs) from protein-coding transcripts is essential for functional genomics research.
• Numerous computational methods have been developed for lncRNA prediction, but a comprehensive review and accessible tool are lacking.

Purpose of the Study:

• To systematically review existing computational methods for lncRNA prediction.
• To develop a user-friendly computational tool for applying state-of-the-art lncRNA prediction methods.

Main Methods:

• Literature review of databases and features used in lncRNA prediction models.
• Summary of computational approaches including binary classifiers, deep learning, and ensemble learning.
• Development of a Python package, ezLncPred, offering a command-line interface for nine advanced prediction methods.

Main Results:

• Identification and categorization of key databases and features crucial for lncRNA prediction model development.
• Comprehensive overview of current state-of-the-art computational prediction strategies.
• Successful implementation of ezLncPred, a practical tool enabling easy access to multiple advanced lncRNA prediction algorithms.

Conclusions:

• The ezLncPred package addresses the need for a user-friendly tool to apply sophisticated lncRNA prediction methods.
• Further research is needed to overcome current challenges in lncRNA prediction and explore future directions.