Gene regulatory network inference based on novel ensemble method.

Area of Science:

• Computational biology
• Bioinformatics
• Systems biology

Background:

• Gene regulatory networks (GRNs) are crucial for understanding gene function, disease development, and cancer.
• Accurate GRN identification is essential for advancing biological and medical research.

Purpose of the Study:

• To develop an advanced ensemble method for improved GRN identification accuracy.
• To enhance the understanding of gene function and its role in diseases like cancer.

Main Methods:

• An ensemble method combining 13 basic classification algorithms (e.g., Random Forest, XGBoost, SVM) with a flexible neural tree (FNT).
• A hybrid evolutionary algorithm integrating gene programming and particle swarm optimization to optimize the FNT model.
• Validation using simulation datasets and real single-cell RNA-seq data.

Main Results:

• The proposed ensemble method demonstrated superior performance compared to 13 supervised, 7 unsupervised, and 4 single-cell-specific GRN identification algorithms.
• Performance was evaluated using Area Under the Receiver Operating Characteristic Curve (AUC-ROC), Area Under the Precision-Recall Curve (AUC-PR), and F1 scores.
• The ensemble approach achieved higher accuracy in GRN identification across diverse datasets.

Conclusions:

• The developed ensemble method offers a significant advancement in GRN identification accuracy.
• This improved accuracy facilitates a deeper understanding of gene regulatory mechanisms and their implications in diseases.
• The study highlights the potential of ensemble learning and evolutionary algorithms in computational biology.