PALM-H3

PALM-H3 PALM

Peer-reviewed Prospective flag Direct antibody flag

Summary card

Поле	Значение
Год / version	2024 / NR
Last verified	2026-08-13
Что генерирует	CDR-H3 sequence; grafted antibody candidates
Форматы	Antibody variable region candidates; validated on SARS-CoV-2 antibodies
De novo или optimization	De novo CDR-H3 on existing framework
DOI	10.1038/s41467-024-50903-y
Paper	https://doi.org/10.1038/s41467-024-50903-y
GitHub / repo	https://github.com/TencentAILabHealthcare/PALM
Service / docs	NR
Лицензия	Custom repository license; verify commercial use
Доступность / commercial-use	Code + Zenodo models/data

Задача и generated object

Used in these tasks: Генерация CDR sequence, Antigen-conditioned antibody design

Part of these workflows: Antigen-conditioned CDR-H3 generation (PALM-H3)

Поле	Значение
Required input	Antigen/epitope sequence and antibody context/framework
Conditioning	CDR-H3 length/context; antigen epitope sequence
Fixed	NR
Editable	NR
Generated	CDR-H3 sequence; grafted antibody candidates
Single mutations	No
Combinatorial mutations	Full CDR-H3
Multi-objective	No explicit multi-objective optimizer

Architecture and operating principle

Author-reported (matrix)

Поле	Значение
Base architecture	NR
Model type	1 + 5 Antibody sequence and target-conditioned generator
Representation	NR
Generation unit	CDR-H3 sequence; grafted antibody candidates
Conditioning	CDR-H3 length/context; antigen epitope sequence
Output	CDR-H3 sequence; grafted antibody candidates
Sampling	NR
Internal scorer	A2binder specificity/affinity predictor
Structure generation	No native 3D generator; structure modeled downstream

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

CDR-H3 length/context; antigen epitope sequence

Generation mechanism

Antigen encoder + autoregressive antibody decoder; A2binder predictor for ranking

Sampling: NR

Training objectives

Losses: NR

Mathematical details: NR

Integrated evaluation

A2binder specificity/affinity predictor

Architectural limitations

Single antigen family in prospective validation.
sequence-only CDR generation.
internal affinity predictor is not affinity improvement.

Uses these architecture components: Protein language models, Autoregressive generation

Generator versus evaluation

Component	Role	Architecture class	Integrated or external	Required or optional	Influences generation or post hoc	Runtime included or excluded
PALM-H3	primary_generator	Antigen encoder + autoregressive antibody decoder; A2binder predictor for ranking	integrated	required	generation	included
A2binder specificity/affinity predictor	integrated_scorer	NR	integrated	optional	post_hoc	unknown
Structure prediction/docking and binding/neutralization 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	3–4 prospective binding and functional validation
Retrospective	NR
Prospective binding	Binding and neutralization reported; predicted affinity outputs are not the experimental result
Prospective function	Binding and neutralization reported; predicted affinity outputs are not the experimental result
Developability	NR
Tested design count	Generated antibodies tested against SARS-CoV-2 variants; exact final count: NR here
Success criterion	Binding and neutralization reported; predicted affinity outputs are not the experimental result
Independent validation	NR
Summary	Binding and neutralization reported; predicted affinity outputs are not the experimental result

Limitations and failure modes

Single antigen family in prospective validation.
sequence-only CDR generation.
internal affinity predictor is not affinity improvement.

Deployment: Limited antibody pilot

Reproducibility, license, commercial use

Поле	Значение
Repo	https://github.com/TencentAILabHealthcare/PALM
License	Custom repository license; verify commercial use
Commercial-use status	Code + Zenodo models/data

Alternative tools: NR

Complementary: NR

Related: NR

Not directly comparable with

AlphaProteo, BindCraft, BoltzGen, FrameDiff, RFdiffusion, RFdiffusion All-Atom (RFdiffusionAA)

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

xTrimoPGLM-based antigen-specific antibody CDR-H3 design (PALM-H3)

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

Review queue

Bibliographic title must be verified against the PDF.

Детальный source-grounded разбор

1. Что именно делает инструмент

[MATRIX]Вход: Antigen/epitope sequence and antibody context/framework

[MATRIX]Generated object: CDR-H3 sequence; grafted antibody candidates

[MATRIX]Фиксируется: NR. Изменяется: NR.

[MATRIX]De novo или parent-based: De novo CDR-H3 on existing framework.

2. Архитектура в одном абзаце

[MATRIX]Antigen encoder + autoregressive antibody decoder; A2binder predictor for ranking

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: CDR-H3 sequence; grafted antibody candidates. Structure generation: No native 3D generator; structure modeled downstream.

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): Binding and neutralization reported; predicted affinity outputs are not the experimental result

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): 3–4 prospective binding and functional validation. Prospective binding: Binding and neutralization reported; predicted affinity outputs are not the experimental result. Function: Binding and neutralization reported; predicted affinity outputs are not the experimental result. Developability: NR.

Single antigen family in prospective validation.
sequence-only CDR generation.
internal affinity predictor is not affinity improvement.

15. Reproducibility

NRРепозиторий: https://github.com/TencentAILabHealthcare/PALM. Лицензия: Custom repository license; verify commercial use. Commit: NR, пока страница не имеет отдельного analysis YAML.

Модульная схема

Модуль	Класс модели	Вход	Выход	Обучается или frozen	Generator или evaluator
PALM-H3	Antigen encoder + autoregressive antibody decoder; A2binder predictor for ranking	NR	NR	NR	primary_generator
A2binder specificity/affinity predictor	NR	NR	NR	NR	integrated_scorer
Structure prediction/docking and binding/neutralization 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
PALM-H3	primary_generator	integrated	required	generation
A2binder specificity/affinity predictor	integrated_scorer	integrated	optional	post_hoc
Structure prediction/docking and binding/neutralization 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
Binding and neutralization reported; predicted affinity outputs are not the experimental result	Antigen encoder + autoregressive antibody decoder; A2binder predictor for ranking	5. prospective functional validation	Generated antibodies tested against SARS-CoV-2 variants; exact final count: NR here	NR	Binding and neutralization reported; predicted affinity outputs are not the experimental result	[MATRIX] landscape matrix 2026-08-13	Не читать как полный разбор primary paper.