Research map / Probabilistic and scientific modeling / Molecules and proteins
h-MINT: Modeling Pocket-Ligand Binding with Hierarchical Molecular Interaction Network
Yanru Qu, Yijie Zhang, Wenjuan Tan, Xiangzhe Kong, Xiangxin Zhou, Chaoran Cheng, Mathieu Blanchette, Jiaxuan You, Ge Liu
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 |
|---|---|---|---|
| use OverlapBPE tokenization instead of non-overlapping PS tokenization in h-MINT (ablation) | h-MINT · LBA | Pearson: 0.633 -> 0.660 | Improved |
| use h-MINT architecture instead of GET with the same PS tokenizer (ablation) | PS tokenization · LBA | Pearson: 0.631 -> 0.633 | No clear effect |
| add chirality-preserving tokens to OverlapBPE (ablation) | OverlapBPE with XGBoost · PubChem HTS (8 assays) | logAUC average rank: 4.000 -> 2.625 | Improved |
| add auxiliary MSE regression loss to LigUnity (ablation) | LigUnity · DUD-E | AUC (%): 81.69 -> 82.57 | Improved |
| add auxiliary MSE regression loss to LigUnity (ablation) | LigUnity · LIT-PCBA | BEDROC (%): 4.34 -> 5.64 | Improved |
| replace LigUnity+MSE with h-MINT adapter (ablation) | LigUnity+MSE · LIT-PCBA | BEDROC (%): 5.64 -> 6.21 | Improved |
| replace LigUnity+MSE with h-MINT adapter (ablation) | LigUnity+MSE · DUD-E | BEDROC (%): 47.58 -> 47.65 | No clear effect |
| increase OverlapBPE vocabulary size from 98 to 212 (ablation) | h-MINT · PDBBind | Pearson: 0.618 -> 0.640 | Improved |
| increase OverlapBPE vocabulary size from 212 to 294 (ablation) | h-MINT · PDBBind | Pearson: 0.640 -> 0.622 | Hurt |
| use h-MINT with OverlapBPE instead of GET-PS | GET-PS · PDBBind | Pearson: 0.601 -> 0.640 | Improved |
| use h-MINT with OverlapBPE instead of PS tokenization (case study) | PS tokenization · LBA case study | error: 0.67 -> 0.56 | Improved |
| use h-MINT adapter instead of LigUnity | LigUnity · LIT-PCBA | AUC (%): 57.61 -> 57.77 | No clear effect |
Closest papers by meaning
- Self-supervised Pocket Pretraining via Protein Fragment-Surroundings Alignment (ICLR 2024 · similarity 0.87)
- ShEPhERD: Diffusing shape, electrostatics, and pharmacophores for bioisosteric drug design (ICLR 2025 · similarity 0.87)
- AANet: Virtual Screening under Structural Uncertainty via Alignment and Aggregation (NeurIPS 2025 · similarity 0.87)
- CBGBench: Fill in the Blank of Protein-Molecule Complex Binding Graph (ICLR 2025 · similarity 0.87)
- Neural Atoms: Propagating Long-range Interaction in Molecular Graphs through Efficient Communication Channel (ICLR 2024 · similarity 0.86)
- UniSite: The First Cross-Structure Dataset and Learning Framework for End-to-End Ligand Binding Site Detection (NeurIPS 2025 · similarity 0.86)
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.
