AbDiffuser

Preprint Prospective flag Direct antibody flag

Summary card

Поле	Значение
Год / version	2023/2024 / NR
Last verified	2026-08-13
Что генерирует	Full-atom antibody sequence + structure
Форматы	Antibody variable domains; HER2 campaign
De novo или optimization	De novo-like antibody generation around learned/reference distribution
DOI	10.48550/arXiv.2308.05027
Paper	https://arxiv.org/abs/2308.05027
GitHub / repo	NR
Service / docs	NR
Лицензия	NR
Доступность / commercial-use	No verified turnkey public code

Задача и generated object

Used in these tasks: Joint CDR sequence–structure design, De novo antibody design

Part of these workflows: Antigen-conditioned CDR co-design

Поле	Значение
Required input	Reference/framework context; generation settings
Conditioning	Antibody priors; sequence-length variation
Fixed	NR
Editable	NR
Generated	Full-atom antibody sequence + structure
Single mutations	Possible
Combinatorial mutations	Yes
Multi-objective	No explicit Pareto optimizer

Architecture and operating principle

Author-reported (matrix)

Поле	Значение
Base architecture	NR
Model type	4 Joint antibody sequence–structure generator
Representation	NR
Generation unit	Full-atom antibody sequence + structure
Conditioning	Antibody priors; sequence-length variation
Output	Full-atom antibody sequence + structure
Sampling	NR
Internal scorer	Physics/domain priors and computational selection
Structure generation	Generates full-atom structure

Input representation

Поле	Значение
Node features	NR
Edge / pair features	NR

Encoder и decoder

Поле	Значение
Encoder	NR
Decoder	NR

Latent or generative state

Conditioning mechanism

Antibody priors; sequence-length variation

Generation mechanism

Equivariant physics-informed diffusion with aligned antibody priors

Sampling: NR

Training objectives

Losses: NR

Mathematical details: NR

Integrated evaluation

Physics/domain priors and computational selection

Architectural limitations

Preprint and no verified implementation.
selected-design success rate is selection-conditional and target-specific.

Uses these architecture components: Joint sequence–structure diffusion, SE(3)-equivariant networks, Joint sequence–structure generation

Generator versus evaluation

Component	Role	Architecture class	Integrated or external	Required or optional	Influences generation or post hoc	Runtime included or excluded
AbDiffuser	primary_generator	Equivariant physics-informed diffusion with aligned antibody priors	integrated	required	generation	included
Physics/domain priors and computational selection	integrated_scorer	NR	integrated	optional	post_hoc	unknown
Generates full-atom structure	integrated_structure_predictor	NR	integrated	optional	post_hoc	included
Independent complex prediction and binding/developability assays	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	NR
Preprocessing	NR
Negative examples	NR
Fine-tuning	NR
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	NR
CPU	NR
GPU	NR
VRAM	NR
RAM	NR
Disk	NR
Databases	NR

Experimental evidence

Поле	Значение
Evidence class	3 prospective binding + expression validation
Retrospective	NR
Prospective binding	Prospective HER2 binding and expression; publication remains preprint
Prospective function	NR
Developability	NR
Tested design count	16 HER2 antibody designs expressed; 57.1% of selected designs reported as tight binders
Success criterion	Prospective HER2 binding and expression; publication remains preprint
Independent validation	NR
Summary	Prospective HER2 binding and expression; publication remains preprint

Limitations and failure modes

Preprint and no verified implementation.
selected-design success rate is selection-conditional and target-specific.

Deployment: Frontier / watchlist

Reproducibility, license, commercial use

Поле	Значение
Repo	NR
License	NR
Commercial-use status	No verified turnkey public code

Alternative tools: NR

Complementary: NR

Related: NR

Not directly comparable with

AntiFold, Efficient evolution with ESM-1v, ESM-IF1, LigandMPNN, ProteinMPNN, Structural evolution (ESM-IF1 complex)

Сравнение допустимо только внутри совместимых категорий. Runtime и quality AntiFold vs RFantibody нельзя читать как одну шкалу: разные starting points и generated objects.

Depends on these external modules

Primary paper и полный список источников

AbDiffuser: full-atom generation of in vitro functioning antibodies

Source matrix: landscape xlsx, last verified 2026-08-13.

Review queue

Bibliographic title must be verified against the PDF.
Official repository/service not verified.

Детальный source-grounded разбор

1. Что именно делает инструмент

[MATRIX]Вход: Reference/framework context; generation settings

[MATRIX]Generated object: Full-atom antibody sequence + structure

[MATRIX]Фиксируется: NR. Изменяется: NR.

[MATRIX]De novo или parent-based: De novo-like antibody generation around learned/reference distribution.

2. Архитектура в одном абзаце

[MATRIX]Equivariant physics-informed diffusion with aligned antibody priors

4. Математическая модель

NRТочные уравнения, losses и sampling solver будут внесены после чтения paper/code. Не подставляйте родовой UCB/transformer/diffusion, если он не подтверждён для этого инструмента.

Поля NR: conditional probability, update equations, sampling algorithm

5. Физическое устройство состояния

[MATRIX]Seed generated state / output: Full-atom antibody sequence + structure. Structure generation: Generates full-atom structure.

6. Физико-химическая интерпретация

[INTERPRETATION]Координаты, frames, PLM embeddings или residue tokens не являются расчётом ΔG, электростатики или растворителя, пока это не сделано внешней energy function. Likelihood / pLDDT / docking score ≠ экспериментальный binding.

7. Training

[MATRIX]Данные: NR. Fine-tuning: NR. Negative examples: NR.

8. Inference

[MATRIX]Sampling: NR. Generation steps: NR. Runtime boundary: NR.

10. Почему, по мнению авторов, метод эффективен

[MATRIX]Author-reported summary (матрица): Prospective HER2 binding and expression; publication remains preprint

NRCausal support не классифицирован, пока не прочитаны ablation и experimental sections.

12. Ablation analysis

NRAblation в этом выпуске не разобрана.

13. What the model does not represent

Явный растворитель и ионная сила — обычно отсутствуют в генераторе.
Протонирование, гликозилирование, кофакторы — обычно отсутствуют, если не оговорено.
Конформационная энтропия и долгая динамика — не моделируются одним forward pass.
Экспрессия, агрегация, иммуногенность — не равны likelihood генератора.
Истинная binding free energy — не равна internal score.

14. Evidence and limitations

[MATRIX]Evidence class (матрица): 3 prospective binding + expression validation. Prospective binding: Prospective HER2 binding and expression; publication remains preprint. Function: NR. Developability: NR.

Preprint and no verified implementation.
selected-design success rate is selection-conditional and target-specific.

15. Reproducibility

NRРепозиторий: NR. Лицензия: NR. Commit: NR, пока страница не имеет отдельного analysis YAML.

Модульная схема

Модуль	Класс модели	Вход	Выход	Обучается или frozen	Generator или evaluator
AbDiffuser	Equivariant physics-informed diffusion with aligned antibody priors	NR	NR	NR	primary_generator
Physics/domain priors and computational selection	NR	NR	NR	NR	integrated_scorer
Generates full-atom structure	NR	NR	NR	NR	integrated_structure_predictor
Independent complex prediction and binding/developability assays	NR	NR	NR	NR	mandatory_external_dependency
Experimental binding/function assays	NR	NR	NR	NR	experimental_validation

Generator versus evaluator

Компонент	Роль	Integrated или external	Required или optional	Влияет на generation или только ранжирует после
AbDiffuser	primary_generator	integrated	required	generation
Physics/domain priors and computational selection	integrated_scorer	integrated	optional	post_hoc
Generates full-atom structure	integrated_structure_predictor	integrated	optional	post_hoc
Independent complex prediction and binding/developability assays	mandatory_external_dependency	external	required	post_hoc
Experimental binding/function assays	experimental_validation	external	required	post_hoc

Claim–evidence matrix

Утверждение	Предполагаемый механизм	Тип evidence	Датасет / эксперимент	Comparator	Метрика / результат	Источник	Оговорка
Prospective HER2 binding and expression; publication remains preprint	Equivariant physics-informed diffusion with aligned antibody priors	5. prospective functional validation	16 HER2 antibody designs expressed; 57.1% of selected designs reported as tight binders	NR	Prospective HER2 binding and expression; publication remains preprint	[MATRIX] landscape matrix 2026-08-13	Не читать как полный разбор primary paper.