GNN2Pfam: Integrating protein sequence and structure with graph neural networks for Pfam domain annotation

Key Points

• Integrating protein sequence and structure improves Pfam domain annotation accuracy significantly.
• The method achieves high performance, with an accuracy rate exceeding 85% across multiple datasets.
• Combining graph neural networks with sequence and structural data enhances traditional annotation methods.
• These findings support using neural networks for improved protein classification in bioinformatics.