LMFE: A Novel Method for Predicting Plant LncRNA Based on Multi-Feature Fusion and Ensemble Learning

Area of Science:

• Genomics
• Computational Biology
• Plant Science

Background:

• Long non-coding RNAs (lncRNAs) are critical regulators of plant traits and disease resistance.
• Plant lncRNA research lags behind animal/human studies due to data scarcity and genomic complexity.
• Accurate lncRNA prediction is vital for advancing plant biology and experimental guidance.

Purpose of the Study:

• To develop an effective computational method for predicting plant lncRNAs.
• To classify transcribed RNA sequences as lncRNAs or messenger RNAs (mRNAs) using multi-feature analysis.
• To address challenges in plant lncRNA identification, including data imbalance and genomic diversity.

Main Methods:

• Proposed the lncRNA multi-feature-fusion ensemble learning (LMFE) approach.
• Integrated 100-dimensional features (biological, sequence, structure-based).
• Employed XGBoost ensemble learning and synthetic minority oversampling technique (SMOTE) for imbalanced data.

Main Results:

• LMFE achieved high accuracy (99.42%) and F1-score (0.99) on benchmark datasets.
• Demonstrated robust cross-species performance (accuracy 89.30%-99.81%).
• Outperformed state-of-the-art methods (CPC2, PLEKv2) on independent datasets with minimal impact from redundant features.

Conclusions:

• LMFE offers a highly accurate and generalizable solution for plant lncRNA prediction.
• The multi-feature fusion and ensemble learning strategy enhances predictive power.
• Future work should focus on improving performance across diverse plant genomes.