LncFinder: an integrated platform for long non-coding RNA identification utilizing sequence intrinsic composition,...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Accurate identification of long non-coding RNAs (lncRNAs) is crucial for biological research.
• Existing machine learning tools often rely solely on sequence features, leading to performance instability across species.
• Current tools lack user customization for feature integration and model retraining.

Purpose of the Study:

• To develop an integrated platform, LncFinder, for enhanced lncRNA identification.
• To design and evaluate novel, heterologous features for improved lncRNA prediction.
• To provide researchers with a flexible tool for lncRNA analysis and feature exploration.

Main Methods:

• Comprehensive review and evaluation of sequence-intrinsic composition, secondary structure, and physicochemical properties as features.
• Development of LncFinder, an R package and web server, incorporating a novel predictor.
• Utilizing multiple machine learning algorithms for classifier construction and performance evaluation.

Main Results:

• LncFinder demonstrates superior performance compared to state-of-the-art methods across multiple species.
• The integrated features lead to more robust and accurate lncRNA identification.
• The platform facilitates feature extraction, classifier building, and performance assessment.

Conclusions:

• LncFinder significantly advances lncRNA identification, particularly for understudied species.
• The tool's flexibility supports diverse research needs, including lncRNA-protein interactions and evolutionary studies.
• LncFinder is available as an R package and web server to maximize accessibility.