RFantibody
RFantibody
Peer-reviewed Prospective flag Direct antibody flagSummary card
| Поле | Значение |
|---|---|
| Год / version | 2025/2026 / NR |
| Last verified | 2026-08-13 |
| Что генерирует | CDR backbones, CDR sequences, antibody–target complex proposals |
| Форматы | VHH/nanobody and scFv; conversion of selected scFvs to IgG tested |
| De novo или optimization | De novo CDR-mediated binder on a chosen framework |
| DOI | 10.1038/s41586-025-09721-5 |
| Paper | https://doi.org/10.1038/s41586-025-09721-5 |
| GitHub / repo | https://github.com/RosettaCommons/RFantibody |
| Service / docs | NR |
| Лицензия | MIT; free for commercial use |
| Доступность / commercial-use | Code + weights |
Задача и generated object
Used in these tasks: De novo antibody design, De novo VHH / nanobody design, Antigen-conditioned antibody design, Генерация CDR backbone, Joint CDR sequence–structure design
Part of these workflows: Target-conditioned de novo antibody (RFantibody)
| Поле | Значение |
|---|---|
| Required input | Target structure, epitope hotspots, fixed antibody framework, CDR lengths |
| Conditioning | Fixed framework, hotspots, CDR lengths, VHH/scFv format |
| Fixed | Antibody framework and target (except designed CDRs) |
| Editable | CDR backbones and sequences |
| Generated | CDR geometry, CDR sequences, complex proposals |
| Single mutations | No |
| Combinatorial mutations | Yes across CDRs |
| Multi-objective | No explicit multi-objective optimizer |
Architecture and operating principle
Author-reported (matrix + Nature paper status)| Поле | Значение |
|---|---|
| Base architecture | RFdiffusion (Ab) + ProteinMPNN + RF2 |
| Model type | Multi-stage generative workflow |
| Representation | SE(3) frames, then residue graph |
| Generation unit | CDR frames then amino-acid tokens |
| Conditioning | Target, hotspots, fixed framework, CDR lengths |
| Output | Sequence–structure antibody–target proposals |
| Sampling | Diffusion trajectories + MPNN sampling |
| Internal scorer | RF2 confidence / self-consistency |
| Structure generation | Да (CDRs/backbone) |
Input representation
Target structure + epitope hotspots + fixed antibody framework + CDR length contigs. Framework coordinates are clamped; CDRs are generated.
| Поле | Значение |
|---|---|
| Node features | NR |
| Edge / pair features | NR |
Encoder и decoder
Equivariant RoseTTAFold-derived trunk as diffusion denoiser. ProteinMPNN is a separate encoder/decoder on the resulting backbone. RF2 is not a generator.
| Поле | Значение |
|---|---|
| Encoder | RoseTTAFold-derived equivariant denoiser (antibody-fine-tuned) |
| Decoder | Reverse diffusion over frames then MPNN sequence decoder |
Latent or generative state
Continuous noisy 3D frames for designed CDR residues, then discrete sequence tokens.
Conditioning mechanism
Coordinate clamping of framework and target; hotspot residue constraints; contig specification of CDR lengths; VHH vs scFv format choice.
Generation mechanism
1) Зашумляются CDR frames. 2) Forward process — SE(3) noise as in RFdiffusion. 3) Denoiser предсказывает cleaner frames. 4) Число sampling steps — implementation/default, exact default NR here. 5) Framework и target фиксируются. 6) Sequence назначается MPNN после backbone. 7) RF2 фильтрует комплексы.
Sampling: Multiple diffusion trajectories; MPNN samples per backbone; RF2/AF3 ranking
Training objectives
Antibody-fine-tuned diffusion objective on structural data; MPNN CE; RF2 trained as structure/complex predictor. Do not collapse these losses into one 'affinity loss'.
Losses: Diffusion score-matching / frame denoising (RFdiffusion family); MPNN cross-entropy; RF2 structure losses are predictor losses
Mathematical details: NR
Integrated evaluation
RF2 is an integrated complex predictor/filter. AF3 is described as retrospective/recommended external filter. Neither is experimental affinity. Display/SPR remain mandatory for binding claims.
Architectural limitations
| Архитектурная особенность | Возможное ограничение |
|---|---|
| Fixed framework | Нет global Fv redesign. |
| Diffusion with many steps | Высокая inference cost. |
| Hotspot-conditioned generation | Ошибка epitope mapping почти фатальна. |
| Integrated RF2 ranking | Возможна miscalibration вне training distribution. |
| No negative antigen examples as primary objective | Специфичность не доказана генератором. |
Uses these architecture components: Frame diffusion, Inverse folding, SE(3)-equivariant networks
Generator versus evaluation
| Component | Role | Architecture class | Integrated or external | Required or optional | Influences generation or post hoc | Runtime included or excluded |
|---|---|---|---|---|---|---|
| Antibody RFdiffusion + ProteinMPNN | primary_generator | Antibody-fine-tuned RFdiffusion → ProteinMPNN CDR sequence → fine-tuned RoseTTAFold2 filter | integrated | required | generation | included |
| Fine-tuned RF2 self-consistency/confidence; AF3 retrospective filtering | integrated_scorer | NR | integrated | optional | post_hoc | unknown |
| Fine-tuned RoseTTAFold2 included | integrated_structure_predictor | NR | integrated | optional | post_hoc | included |
| AF3 recommended retrospectively; experimental binding/epitope/function tests mandatory | mandatory_external_dependency | NR | external | required | post_hoc | excluded |
| Experimental binding/function assays | experimental_validation | NR | external | required | post_hoc | excluded |
Training data и biases
| Поле | Значение |
|---|---|
| Training data | Antibody-fine-tuned RFdiffusion/RF2 (exact corpora: NR here) |
| Preprocessing | NR |
| Negative examples | NR as an explicit specificity-training set |
| Fine-tuning | Yes, from general RFdiffusion/RF2 toward antibody geometry |
| Label requirements | NR |
| Training hardware | NR |
| Training time | NR |
Inference hardware и runtime
| Поле | Значение |
|---|---|
| Candidate count | NR |
| Batch size | NR |
| Generation steps | NR |
| Recycles | NR |
| Size | NR |
| Reported runtime | NR |
| Runtime boundary | Must separate diffusion, MPNN, RF2, optional AF3, and wet-lab |
| CPU | NR |
| GPU | Required (qualitative; hours/VRAM NR) |
| VRAM | NR |
| RAM | NR |
| Disk | NR |
| Databases | No MSA database required for the diffusion stage (Author-reported workflow character); RF2 details NR |
Experimental evidence
| Поле | Значение |
|---|---|
| Evidence class | 3–4 prospective binding and functional validation |
| Retrospective | NR |
| Prospective binding | Correct epitope and designed geometry for selected structures; TcdB neutralization after maturation; initial affinities can be modest |
| Prospective function | Correct epitope and designed geometry for selected structures; TcdB neutralization after maturation; initial affinities can be modest |
| Developability | NR |
| Tested design count | 95-design low-throughput or 9,000-design display campaigns per target; overall target success 0–2%; multiple confirmed VHH/scFv binders |
| Success criterion | Correct epitope and designed geometry for selected structures; TcdB neutralization after maturation; initial affinities can be modest |
| Independent validation | NR |
| Summary | Correct epitope and designed geometry for selected structures; TcdB neutralization after maturation; initial affinities can be modest |
Limitations and failure modes
- Framework fixed.
- target holo conformation/hotspots critical.
- lower hit rates than general miniprotein binder pipelines.
- CDR humanness not optimized.
Deployment: Limited antibody pilot
Reproducibility, license, commercial use
| Поле | Значение |
|---|---|
| Repo | https://github.com/RosettaCommons/RFantibody |
| License | MIT; free for commercial use |
| Commercial-use status | Code + weights |
Related tools
Alternative tools: tFold System, BoltzGen
Complementary: ProteinMPNN, AntiFold
Related: RFdiffusion, ProteinMPNN, tFold System
Not directly comparable with
AlphaProteo, BindCraft, BoltzGen, Efficient evolution with ESM-1v, Structural evolution (ESM-IF1 complex)Depends on these external modules
- RoseTTAFold2 / antibody-fine-tuned RF2
- AlphaFold3
- Experimental binding/function assays
- proteinmpnn
Related glossary terms
Diffusion model, Inverse folding, Epitope / hotspot, Framework, CDR, pLDDT / ipTM / PAEPrimary paper и полный список источников
Atomically accurate de novo design of antibodies with RFantibodySource matrix: landscape xlsx, last verified 2026-08-13.
Review queue
- Bibliographic title must be verified against the PDF.
Детальный source-grounded разбор
Недоступные или неполные источники
| Источник | Почему недоступен | Что использовано вместо |
|---|---|---|
| Nature Methods/SI complete | HTML частично; SI не выгружен | Narrative Results pipeline; catalog numbers from matrix cross-check |
| bioRxiv 10.1101/2024.03.14.585103 | HTTP 429 | Nature version |
| RosettaCommons/RFantibody commit SHA | API timeout | URL репозитория без SHA |
RFantibody — многостадийный workflow: antibody-fine-tuned RFdiffusion (CDR frames) → ProteinMPNN (sequence) → antibody-fine-tuned RoseTTAFold2 (filter). AF3 в исследовании — retrospective filter. Binding устанавливается assay/cryo-EM, не pLDDT.
Происхождение источников
| Метка | Источник | Ссылка | Примечание |
|---|---|---|---|
| [PAPER] | Bennett, Watson et al. Nature 10.1038/s41586-025-09721-5 | https://doi.org/10.1038/s41586-025-09721-5 | NR |
| [DOC] | https://github.com/RosettaCommons/RFantibody MIT | https://github.com/RosettaCommons/RFantibody | NR |
| Поле | Значение |
|---|---|
| Репозиторий | https://github.com/RosettaCommons/RFantibody |
| Commit | NR (timeout) |
| Ветка / tag | main |
| Лицензия | MIT (каталог YAML) |
Проверенные файлы реализации:
NR — listing не получен в этом проходе
1. Что именно делает инструмент
[PAPER]Вход: структура мишени, epitope hotspots, фиксированный antibody framework, длины CDR / contig, формат VHH или scFv. Выход: CDR backbones + sequences + комплекс-предложения. Фиксируется: framework и target (кроме designed CDRs). De novo CDR-mediated binder, не parent affinity maturation AL.
2. Архитектура в одном абзаце
[PAPER]Не одна сеть. (1) SE(3) diffusion denoiser на базе RFdiffusion, fine-tuned на antibody geometry, генерирует CDR frames при clamped FW/target. (2) ProteinMPNN назначает CDR sequences на полученный backbone. (3) Fine-tuned RF2 оценивает комплекс (self-consistency/confidence) и отфильтровывает кандидатов. AF3 — дополнительный retrospective ranker в статье, не generator.
4. Математическая модель
NRТочные score-matching уравнения Ab-RFdiffusion в этом выпуске NR (Methods). Классово: reverse SE(3) на designed frames. MPNN: $p(s\mid x_{\mathrm{bb}})$. RF2: structure/complex prediction loss, не $p(K_d)$.
5. Физическое устройство состояния
[PAPER]Стадия 1: noisy/clean residue frames CDRs. Стадия 2: amino-acid tokens. Стадия 3: predicted complex coordinates + confidence maps. Experimental fitness — только после display/SPR/cryo-EM.
6. Физико-химическая интерпретация
[PAPER]Steric complementarity может отражаться в learned geometry + later RF2. H-bonds/electrostatics/solvation не explicit в diffusion. RF2/AF3 confidence — статистика структур, не ΔG. OrthoRep maturation — экспериментальная эволюция после слабых начальных Kd.
7. Training
NRAntibody fine-tune RFdiffusion и RF2; MPNN стандартный sequence recovery. Точные датасеты, фильтры, hardware — NR без Methods. Negative antigens как specificity objective — не описаны как главный loss.
8. Inference
[PAPER]Кампании статьи: порядка 95 designs low-throughput или ~9000 display per target (цифры каталога/Nature narrative; уточнять по paper figures). Hit rate overall target success 0–2% (каталог). GPU hours/VRAM NR. Отделять diffusion wall-time от MPNN, RF2 recycles, AF3, wet-lab.
10. Почему, по мнению авторов, метод эффективен
[PAPER]Mechanism: antibody-specific fine-tune diffusion на CDR; hotspot-conditioned epitope; MPNN для sequences; RF2 filter для self-consistent complexes; experimental display/SPR/cryo-EM.
[PAPER]Claim: atomically accurate de novo antibodies (VHH/scFv); некоторые нейтрализующие после maturation (TcdB). Initial affinities often modest (tens–hundreds nM).
[INTERPRETATION]Causal support: prospective binding/structure — да для selected designs. Ablation «зачем fine-tune» vs vanilla RFdiffusion — нужно SI; здесь не утверждать цифры. RF2 ranking не доказан как причина binding, только как filter перед экспериментом.
12. Ablation analysis
NRПолные ablation tables Methods/SI в этом проходе не извлечены (NR). Не заполнять вымышленными процентами.
13. What the model does not represent
- Оптимизация humanness CDR
- Явный растворитель, гликаны, Fc-эффектор (если не моделируется отдельно после IgG conversion)
- Конформационная энтропия
- Истинная ΔG
- Гарантия developability/manufacturability
- Free framework redesign
14. Evidence and limitations
[PAPER]Prospective: binders confirmed; cryo-EM для selected; hit rates modest vs miniprotein binder pipelines (authors). Framework fixed. Target holo conformation критичен. Independent validation вне авторов — NR здесь.
15. Reproducibility
NRMIT, GitHub RosettaCommons/RFantibody, weights claimed in catalog. Commit SHA NR (timeout). Inference example не прогонялся в этом выпуске.
Физико-химический смысл
| Признак или операция модели | Физико-химическая интерпретация | Прямо / косвенно / внешне | Важное ограничение |
|---|---|---|---|
| Hotspots | epitope residues | conditioning | user-defined, can be wrong |
| CDR frames | loop backbone | generated | side chains later |
| RF2 PAE/pLDDT-like confidence | predictor self-consistency | evaluator | ≠ affinity |
Модульная схема
| Модуль | Класс модели | Вход | Выход | Обучается или frozen | Generator или evaluator |
|---|---|---|---|---|---|
| Ab-finetuned RFdiffusion | SE(3) frame diffusion | target, hotspots, fixed FW, CDR contig | CDR backbone frames | fine-tuned generator | generator |
| ProteinMPNN | inverse folding | backbone | CDR sequences | typically frozen pretrained MPNN | integrated generator stage |
| Ab-finetuned RoseTTAFold2 | structure/complex predictor | sequence+complex hypothesis | coordinates + confidence | fine-tuned predictor | evaluator / filter |
| AlphaFold3 (study) | complex predictor | sequences | models/confidence | external | retrospective evaluator |
| Display / SPR / cryo-EM / OrthoRep | experiment | designs | binding, structure, matured Kd | n/a | experimental validation |
Generator versus evaluator
| Компонент | Роль | Integrated или external | Required или optional | Влияет на generation или только ранжирует после |
|---|---|---|---|---|
| RFdiffusion (Ab) | backbone generator | integrated | required | generation |
| ProteinMPNN | sequence generator | integrated | required | generation |
| RF2 | complex predictor / ranker | integrated | used as filter | post_hoc (does not denoise frames) |
| AF3 | external complex predictor | external | optional/retrospective | post_hoc |
| Assays | experimental | external | required for binding claims | not generation |
Claim–evidence matrix
| Утверждение | Предполагаемый механизм | Тип evidence | Датасет / эксперимент | Comparator | Метрика / результат | Источник | Оговорка |
|---|---|---|---|---|---|---|---|
| De novo VHH/scFv binders to specified epitopes can be generated | Ab-RFdiffusion + MPNN + filters + experiment | 4. prospective binding validation | multiple targets; 95 or ~9000 designs/campaign (catalog) | undesigned / failed targets; miniprotein pipelines (qualitative) | binding, cryo-EM epitope; hit rate 0–2% overall (catalog) | [PAPER] Nature Results (HTML narrative) | Methods numbers need SI; modest initial Kd |
| Designed geometry matches experimental structures for selected cases | frame diffusion + filtering | 4. prospective binding validation | cryo-EM subset | design model | structural agreement (details SI NR) | [PAPER] Nature | Selected successes, not all trajectories |
| Neutralization after OrthoRep maturation (TcdB) | experimental evolution on designed starting points | 5. prospective functional validation | TcdB campaign | parent designed clone | neutralization / nM affinities after maturation | [PAPER] Nature | Maturation is not the neural generator |
| RF2/AF3 confidence equals affinity | — | 1. только architectural rationale | n/a | n/a | false | [INTERPRETATION] scope rule | Confidence ≠ Kd |
Различия между статьёй и реализацией
| Версия статьи | Commit репозитория | Checkpoint / model | Файл | Различие | Практическое следствие |
|---|---|---|---|---|---|
| Nature 10.1038/s41586-025-09721-5 | NR | NR checkpoints this pass | NR | Реализация не сверена построчно с Methods | Inference hyperparameters брать из repo configs после clone, не из этой страницы |
Схемы information flow
flowchart TD
subgraph gen["Generative stages"]
RF["RFdiffusion Ab — CDR frames"]
MP["ProteinMPNN — sequence"]
end
subgraph eval["Integrated / external evaluators"]
R2["RF2 filter"]
AF["AF3 retrospective"]
end
subgraph exp["Experiment"]
D["Display/SPR"]
C["cryo-EM"]
O["OrthoRep"]
end
RF --> MP --> R2 --> D
R2 --> AF
D --> C
D --> O
Источники
- [1] Bennett, Watson et al. Atomically accurate de novo design of antibodies with RFdiffusion. Nature. DOI 10.1038/s41586-025-09721-5 · DOI · источник
- [2] RosettaCommons/RFantibody · источник
- [3] Dauparas et al. ProteinMPNN. Science 2022. Inverse-folding stage. · DOI
Нерешённые научные вопросы
- Commit SHA, default diffusion steps, RF2 recycle count
- Full ablation of antibody fine-tune
- Developability panel beyond selected IgG conversion