Research map / Probabilistic and scientific modeling / Molecules and proteins
GP-MoLFormer-Sim: Test Time Molecular Optimization Through Contextual Similarity Guidance
Jiří Navrátil, Jarret Ross, Payel Das, Youssef Mroueh, Samuel C Hoffman, Vijil Chenthamarakshan, Brian Belgodere
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 graph-based crossover (XO) (ablation) | guided generation · PMO 23 tasks configuration ablation | average top-10 AUC: 0.603 -> 0.672 | Improved |
| add random Fourier features (RFF768) alone (ablation) | guided generation · PMO 23 tasks configuration ablation | average top-10 AUC: 0.603 -> 0.597 | Hurt |
| add RFF768 on top of XO (ablation) | GG+XO · PMO 23 tasks configuration ablation | average top-10 AUC: 0.672 -> 0.682 | Improved |
| add diverse guides (DIV) on top of XO (ablation) | GG+XO · PMO 23 tasks configuration ablation | average top-10 AUC: 0.672 -> 0.678 | Improved |
| add DIV on top of RFF768+XO (ablation) | GG+RFF768+XO · PMO 23 tasks configuration ablation | average top-10 AUC: 0.682 -> 0.690 | Improved |
| redact molecule name from MOLLEO prompt (thiothixene) (ablation) | MOLLEO GPT-4.1-mini · PMO thiothixene_rediscovery | top-10 AUC: 0.692 -> 0.462 | Hurt |
| redact molecule name from MOLLEO prompt (mestranol) (ablation) | MOLLEO GPT-4.1-mini · PMO mestranol_similarity | top-10 AUC: 0.983 -> 0.644 | Hurt |
| give GPMFS+GA access to target SMILES in the oracle | GPMFS+GA · PMO (tasks with known target) | average top-10 AUC: 0.662 -> 0.737 | Improved |
| use GPMFS+GA instead of MOLLEO (redacted) in black-box setting | MOLLEO GPT-4.1-mini redacted · PMO thiothixene_rediscovery | top-10 AUC: 0.462 -> 0.504 | Improved |
| use GPMFS+GA instead of MOLLEO GPT-4 with name in prompt | MOLLEO (GPT-4) · PMO benchmark (23 tasks, black-box oracle) | average top-10 AUC: 0.777 -> 0.662 | Hurt |
| use GP-MoLFormer-Sim guidance instead of random generations (top-10000) | random generation · Trypsin inhibitor similarity | mean Tanimoto similarity: 0.225 -> 0.573 | Improved |
| use GP-MoLFormer-Sim guidance instead of RL-tuned S model (top-10000) | S model · Trypsin inhibitor similarity | mean Tanimoto similarity: 0.439 -> 0.573 | Improved |
Closest papers by meaning
- MolTailor: Tailoring Chemical Molecular Representation to Specific Tasks via Text Prompts (AAAI 2024 · similarity 0.90)
- Domain-Agnostic Molecular Generation with Chemical Feedback (ICLR 2024 · similarity 0.89)
- Efficient Evolutionary Search Over Chemical Space with Large Language Models (ICLR 2025 · similarity 0.88)
- Text-Guided Molecule Generation with Diffusion Language Model (AAAI 2024 · similarity 0.88)
- GenMol: A Drug Discovery Generalist with Discrete Diffusion (ICML 2025 · similarity 0.88)
- Omni-Mol: Multitask Molecular Model for Any-to-any Modalities (NeurIPS 2025 · similarity 0.87)
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.
