Research map / Probabilistic and scientific modeling / Molecules and proteins
Learning Mutational Semantics
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| lower beta from 1 to 0.25 (reduce influence of grammaticality term) (ablation) | CSCS · headline grammaticality | acceptable headlines (consensus): 174 -> 104 | Hurt |
| use CSCS-mutated headline (beta=1) vs original | headlines · headline grammaticality | acceptable headlines (consensus): 197 -> 174 | No clear effect |
| choose CSCS-proposed mutation instead of semantically closest | mutation selection · headline POS change | median % POS change (NLTK): 0.00% -> 16.7% | Helped |
| choose CSCS-proposed mutation instead of semantically closest | mutation selection · headline WordNet | Wu-Palmer similarity: 0.546 -> 0.235 | Helped |
| add grammaticality to semantic change alone for escape ranking (ablation) | CSCS · Influenza H3 escape | normalized AUC: 0.709 -> 0.771 | Helped |
| add semantic change to grammaticality alone (ablation) | CSCS · Influenza H3 escape | normalized AUC: 0.684 -> 0.771 | Helped |
| add semantic change to grammaticality alone (ablation) | CSCS · Influenza H1 escape | normalized AUC: 0.820 -> 0.834 | Helped |
| add grammaticality to semantic change alone (ablation) | CSCS · HIV Env escape | normalized AUC: 0.622 -> 0.692 | Helped |
| CSCS vs EVcouplings (epistatic) (ablation) | CSCS · Influenza H3 escape | normalized AUC: 0.687 -> 0.771 | Helped |
| CSCS vs EVcouplings (epistatic) (ablation) | CSCS · HIV Env escape | normalized AUC: 0.552 -> 0.692 | Helped |
| replace BiLSTM with densely-connected layers (ablation) | language model · headline LM | test loss: 6.0 -> 7.2 | Hurt |
| remove right context from language model (ablation) | language model · headline LM | test loss: 6.0 -> 6.5 | Hurt |
Closest papers by meaning
- MutaPLM: Protein Language Modeling for Mutation Explanation and Engineering (NeurIPS 2024 · similarity 0.88)
- Knowledge-aware Reinforced Language Models for Protein Directed Evolution (ICML 2024 · similarity 0.85)
- DePLM: Denoising Protein Language Models for Property Optimization (NeurIPS 2024 · similarity 0.84)
- Understanding protein function with a multimodal retrieval-augmented foundation model (NeurIPS 2025 · similarity 0.84)
- CGBench: Benchmarking Language Model Scientific Reasoning for Clinical Genetics Research (NeurIPS 2025 · similarity 0.84)
- LEMON-ZEST: Evolution-Informed Tokenization for Efficient Protein Language Modeling (NeurIPS 2026 · similarity 0.84)
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.