Research map / Probabilistic and scientific modeling / Molecules and proteins
Explicit representation of germline and non-germline residues improves antibody language modeling
Jeonghyeon Kim, Nathaniel Blalock, Ameya Kulkarni, Kensuke Nakamura, Philip Romero
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 |
|---|---|---|---|
| factorize tokens into germline/non-germline types with an Origin-Conditioned LM head (ablation) | antibody language model (ESM2-35M backbone) · GL/NGL linear probe | PR-AUC: 0.377 -> 0.980 | Improved |
| add Origin-Conditioned multi-head instead of simple head on extended vocabulary (ablation) | PRISM · Generative PPL (heavy chain) | pseudo-perplexity: Ablations 2 and 3 ~50 and ~25 heavy-chain PPL vs 1.54 | Improved |
| add unpaired pretraining to the multi-head model (ablation) | PRISM · GL/NGL linear probe | PR-AUC: 0.517 -> 0.980 | Improved |
| use factorization rather than ESM2 initialization with antibody fine-tuning (ablation) | PRISM vs PRISM-less · CR9114 zero-shot affinity | Spearman rho: -0.409 -> +0.393 | Improved |
| use factorization rather than ESM2 fine-tuning (ablation) | PRISM vs PRISM-less · Trastuzumab zero-shot affinity | Spearman rho: 0.176 -> 0.327 | Improved |
| ESM2 fine-tuning without factorization gives strong probe separability but no geometric organization (ablation) | PRISM-less · GL/NGL embedding separability | ARI: PR-AUC 0.921, ARI 0.006 | No clear effect |
| add stop-gradient before conditioning downstream heads on Origin Head (ablation) | PRISM · validation NGL perplexity | NGL perplexity: ~2x improvement | Improved |
| keep loss reweighting (alpha_NGL=3.0) and 53-token vocabulary but drop the architecture (ablation) | PRISM · GL/NGL separability | ARI: 0.162 -> 0.009 | Hurt |
| sample from NGL-only token channel (PRISM-NGL) | PRISM generation · G6.31, CR9114, Trastuzumab | Rosetta interface ddG variants improving | Improved |
| sample from GL-only token channel (PRISM-GL) | PRISM generation · G6.31, CR9114, Trastuzumab | Rosetta stability and CamSol | Improved |
| NGL-biased sampling scored by DMS ridge predictor | PRISM-NGL · CR9114, Trastuzumab | ridge-based predicted binding | Hurt |
| marginalize GL and NGL channels into one score (ablation) | PRISM scoring · binding DMS (Trastuzumab, CR9114) | Spearman rho | Hurt |
Closest papers by meaning
- How to make the most of your masked language model for protein engineering (NeurIPS 2026 · similarity 0.86)
- Antibody Design Using a Score-based Diffusion Model Guided by Evolutionary, Physical and Geometric Constraints (ICML 2024 · similarity 0.86)
- Synergy of GFlowNet and Protein Language Model Makes a Diverse Antibody Designer (AAAI 2025 · similarity 0.86)
- AffinityFlow: Guided Flows for Antibody Affinity Maturation (ICML 2025 · similarity 0.86)
- dyAb: Flow Matching for Flexible Antibody Design with AlphaFold-driven Pre-binding Antigen (AAAI 2025 · similarity 0.86)
- Retrieval Augmented Diffusion Model for Structure-informed Antibody Design and Optimization (ICLR 2025 · 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.
