Virual Oral Presentation Slide by Slide

IEEE/ICACT20260263 Slide.15        [Big Slide]	Oral Presentation
	We compared against four state-of-the-art methods. D-CyPre achieves F1 of 0.669 using graph neural networks but lacks 3D binding analysis - we improve by 26.6%. DEEPCYPs achieves 86-90% AUC but provides no structural context. Multitask models work well but are sequence-only - we add 3D structure integration. General LLM RAG gets NDCG of 0.623 but cannot process protein structures - we improve by 35.9%. Overall, our NDCG@10 of 0.847 represents 29.4% average improvement across all baselines.

[Go to Next Slide]

Select Voice: Virtual Oral Presentation!!