Skip to main content
. 2025 Dec 5;12:1681083. doi: 10.3389/fvets.2025.1681083

Figure 2.

Flowchart illustrating the process for constructing node and edge indexes based on physicochemical properties. It starts with sequence features leading to feature grouping into four groups: AAC, hydrophilicity, secondary structure, and polar charge. This forms graph-formatted data with nodes and features. The process includes GNN model initialization and training (80 percent) for 3000 epochs, followed by testing (20 percent) for model prediction. The outcome provides values of AUC, accuracy, precision, recall, and F1.

Step by step figure of node formation and amino acids features data input of the schematic layout for the GNN model in AVPs prediction.