Research map / Vision-language and multimodal / Cross-modal learning
Breaking the Modality Barrier: Generative Modeling for Accurate Molecule Retrieval from Mass Spectra
Yiwen Zhang, Keyan Ding, Yihang Wu, Xiang Zhuang, Yi Yang, Qiang Zhang, Huajun Chen
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 generative retrieval stage on top of pre-retrieval (ablation) | GLMR · MassSpecGym weight-based | Recall@1: 20.341 -> 64.172 | Improved |
| add pre-retrieval candidates as context for generation (ablation) | GLMR · MassSpecGym weight-based | Recall@1: 41.501 -> 64.172 | Improved |
| add generative retrieval stage (ablation) | GLMR · MassSpecGym formula-based | Recall@1: 46.030 -> 68.478 | Improved |
| add pre-retrieval stage (ablation) | GLMR · MassSpecGym formula-based | Recall@1: 52.968 -> 68.478 | Improved |
| add generative retrieval (ablation) | GLMR · MassSpecGym weight-based | Recall@20: 74.630 -> 78.782 | No clear effect |
| replace JESTR with GLMR | JESTR · MassSpecGym weight-based | Recall@1: 17.617 -> 64.172 | Improved |
| replace JESTR with GLMR zero-shot | JESTR · MassRET-20k formula-based | Recall@1: 7.440 -> 51.141 | Improved |
| use generated-molecule re-ranking to reduce modality gap (ablation) | GLMR · MassSpecGym | modality gap | Improved |
| increase K pre-retrieved molecules beyond 40 (ablation) | GLMR · MassSpecGym weight-based | Recall/MRR | No clear effect |
| use GLMR generator instead of DiffMS for de-novo generation | DiffMS · MassSpecGym test | MCES: 18.45 -> 21.83 | Hurt |
Closest papers by meaning
- From Artificially Real to Real: Leveraging Pseudo Data from Large Language Models for Low-Resource Molecule Discovery (AAAI 2024 · similarity 0.86)
- MADGEN: Mass-Spec attends to De Novo Molecular generation (ICLR 2025 · similarity 0.86)
- Atomas: Hierarchical Adaptive Alignment on Molecule-Text for Unified Molecule Understanding and Generation (ICLR 2025 · similarity 0.86)
- Unraveling Molecular Structure: A Multimodal Spectroscopic Dataset for Chemistry (NeurIPS 2024 · similarity 0.85)
- MS-BART: Unified Modeling of Mass Spectra and Molecules for Structure Elucidation (NeurIPS 2025 · similarity 0.85)
- GLACIER: Rethinking Mass Spectrum Prediction as an Object Detection Problem (NeurIPS 2026 · 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.
