GRA-GCN: Dense Granule Protein Prediction in Apicomplexa Protozoa Through Graph Convolutional Network

Area of Science:

• Parasitology
• Computational Biology
• Bioinformatics

Background:

• Apicomplexa protozoa cause significant farm animal diseases.
• Dense granule proteins (GRAs) secreted by Apicomplexa are key virulence factors.
• Experimental prediction of GRAs is laborious and time-consuming.

Purpose of the Study:

• To develop an efficient computational method for predicting dense granule proteins (GRAs).
• To address the urgent need for accurate GRA prediction in Apicomplexa.
• To provide a tool for understanding parasitic disease mechanisms.

Main Methods:

• A novel computational method, GRA-GCN, was developed using graph convolutional networks.
• The prediction task was framed as a node classification problem in graph theory.
• The k-nearest neighbor algorithm was employed to construct feature graphs for enhanced representation.

Main Results:

• GRA-GCN demonstrated satisfactory performance, validated by 5-fold cross-validation.
• The proposed method outperformed four classic machine learning models.
• GRA-GCN showed superiority over three state-of-the-art prediction models.
• Comprehensive experimental analysis and a case study provided valuable insights.

Conclusions:

• GRA-GCN represents the first computational approach for GRAs prediction in Apicomplexa.
• The developed method offers a more efficient and accurate alternative to experimental prediction.
• The findings contribute to a better understanding of parasitic disease mechanisms and accurate GRA identification.