A genetic algorithm-based weighted ensemble method for predicting transposon-derived piRNAs.

Area of Science:

• Bioinformatics
• Genomics
• Molecular Biology

Background:

• piwi-interacting RNA (piRNA) prediction is crucial for understanding gamete generation.
• Existing machine learning methods for piRNA prediction require further improvement.

Purpose of the Study:

• To develop an improved method for predicting transposon-derived piRNAs.
• To evaluate the proposed method's performance across different species and dataset types.

Main Methods:

• A genetic algorithm-based weighted ensemble method was developed.
• Prediction models were constructed and evaluated using balanced and imbalanced datasets for Human, Mouse, and Drosophila.
• Cross-species prediction capabilities were assessed.

Main Results:

• The proposed method achieved high AUC values (0.932-0.996) on both balanced and imbalanced datasets across species.
• The models demonstrated strong performance in cross-species piRNA identification.
• The method outperformed existing state-of-the-art approaches.

Conclusions:

• The genetic algorithm-based weighted ensemble method offers a promising approach for transposon-derived piRNA prediction.
• The developed models exhibit robust performance and potential for broader applications in piRNA research.
• Source code and datasets are publicly available for reproducibility.