Research map / Probabilistic and scientific modeling / Molecules and proteins
[Re] Learning Unknown from Correlations: Graph Neural Network for Inter-novel-protein Interaction Prediction
Urša Zrimšek
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 |
|---|---|---|---|
| model PPI correlations with a GIN graph over proteins (GNN-PPI vs PIPR) | PIPR · STRING BFS | micro-F1: 55.65 -> 78.37 | Improved |
| graph-based GNN-PPI vs PIPR | PIPR · STRING DFS | micro-F1: 67.45 -> 91.07 | Improved |
| graph-based GNN-PPI vs PIPR | PIPR · SHS27k BFS | micro-F1: 44.48 -> 63.81 | Improved |
| graph-based GNN-PPI vs PIPR on random split | PIPR · SHS148k Random | micro-F1: 90.05 -> 92.26 | Improved |
| replace random split with BFS/DFS split | PIPR · STRING | micro-F1: 94.43 -> 55.65 | Hurt |
| replace random split with BFS split | GNN-PPI · STRING | micro-F1: 95.43 -> 78.37 | Hurt |
| evaluate on trainset-homologous test vs unseen STRING test, Random split | PIPR · SHS148k | micro-F1: 92.42 -> 53.85 | Hurt |
| evaluate with BFS split instead of Random split as generalization proxy | PIPR · SHS148k to STRING | micro-F1: 66.13 vs. 63.74 | Improved |
| build PPI graph from trainset only (GCT) instead of all data (GCA) (ablation) | GNN-PPI · STRING BFS | micro-F1: 78.37 -> 73.23 | Hurt |
| build PPI graph from trainset only (GCT) instead of all data (GCA) (ablation) | GNN-PPI · SHS27k DFS | micro-F1: 74.72 -> 73.42 | Hurt |
| remove PGE graph module (PIE only) (ablation) | GNN-PPI · STRING BFS | micro-F1: 78.37 -> 50.03 | Hurt |
| remove PIE sequence encoder (PGE only, handcrafted features) (ablation) | GNN-PPI · STRING BFS | micro-F1: 78.37 -> 73.81 | Hurt |
Closest papers by meaning
- Dual-Channel Learning Framework for Drug-Drug Interaction Prediction via Relation-Aware Heterogeneous Graph Transformer (AAAI 2024 · similarity 0.83)
- Rethinking Cancer Gene Identification Through Graph Anomaly Analysis (AAAI 2025 · similarity 0.83)
- Revisiting Link Prediction: a data perspective (ICLR 2024 · similarity 0.83)
- MOTIVE: A Drug-Target Interaction Graph For Inductive Link Prediction (NeurIPS 2024 · similarity 0.82)
- Exact Computation of Any-Order Shapley Interactions for Graph Neural Networks (ICLR 2025 · similarity 0.82)
- Knowledge Enhanced Representation Learning for Drug Discovery (AAAI 2024 · similarity 0.81)
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.
