Research map / Vision-language and multimodal / Human and embodied modeling
Enhancing Protein-Protein Interaction Prediction with Hierarchical Motif-based Multimodal Protein Embedding
Zaifei YANG, Samuel Ping-Man Choi, James Kwok
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add hierarchical motif and protein encoders to sequence-only model (ablation) | MMM-PPI · average over 9 splits, transductive | Micro-F1 (%): 83.01 -> 84.67 | Improved |
| add hierarchical motif and protein encoders to structure-only model (ablation) | MMM-PPI · average over 9 splits, transductive | Micro-F1 (%): 83.78 -> 84.86 | Improved |
| add hierarchical motif and protein encoders to function-only model (ablation) | MMM-PPI · average over 9 splits, transductive | Micro-F1 (%): 83.44 -> 84.72 | Improved |
| combine three modalities instead of sequence only with hierarchical encoders (ablation) | MMM-PPI · average over 9 splits, transductive | Micro-F1 (%): 84.67 -> 85.60 | Improved |
| add motif encoder and protein encoder to three-modality model without them (ablation) | MMM-PPI · average over 9 splits, transductive | Micro-F1 (%): 84.01 -> 85.60 | Improved |
| add protein encoder to three-modality model with motif encoder only (ablation) | MMM-PPI · average over 9 splits, transductive | Micro-F1 (%): 84.37 -> 85.60 | Improved |
| add motif encoder to three-modality model with protein encoder only (ablation) | MMM-PPI · average over 9 splits, transductive | Micro-F1 (%): 84.64 -> 85.60 | Improved |
| use three modality-specific HGNNs instead of early fusion (ablation) | MMM-PPI · SHS27k BFS, transductive | Micro-F1 (%): 72.19 -> 79.09 | Improved |
| use three modality-specific HGNNs instead of one HGNN (ablation) | MMM-PPI · SHS27k BFS, transductive | Micro-F1 (%): 75.87 -> 79.09 | Improved |
| use concatenation instead of sum aggregation (ablation) | MMM-PPI · SHS27k BFS, transductive | Micro-F1 (%): 78.28 -> 79.09 | Improved |
| use concatenation instead of MLP aggregation (ablation) | MMM-PPI · SHS148k Random, transductive | Micro-F1 (%): 93.16 -> 93.38 | Improved |
| use all motif lengths instead of only motifs of length 0-10 (ablation) | MMM-PPI · SHS27k BFS, transductive | Micro-F1 (%): 76.30 -> 79.09 | Improved |
Closest papers by meaning
- MAPE-PPI: Towards Effective and Efficient Protein-Protein Interaction Prediction via Microenvironment-Aware Protein Embedding (ICLR 2024 · similarity 0.89)
- PRING: Rethinking Protein-Protein Interaction Prediction from Pairs to Graphs (NeurIPS 2025 · similarity 0.86)
- PPDiff: Diffusing in Hybrid Sequence-Structure Space for Protein-Protein Complex Design (ICML 2025 · similarity 0.86)
- PSC-CPI: Multi-Scale Protein Sequence-Structure Contrasting for Efficient and Generalizable Compound-Protein Interaction Prediction (AAAI 2024 · similarity 0.86)
- ProtCLIP: Function-Informed Protein Multi-Modal Learning (AAAI 2025 · similarity 0.85)
- MMSite: A Multi-modal Framework for the Identification of Active Sites in Proteins (NeurIPS 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
