Humatch

Peer-reviewed

Summary card

Поле	Значение
Год / version	2024 / NR
Last verified	2026-08-13
Что генерирует	Humanized heavy/light variable-domain sequences and mutation set
Форматы	Paired antibody VH/VL
De novo или optimization	Optimizes existing antibody
DOI	10.1080/19420862.2024.2434121
Paper	https://doi.org/10.1080/19420862.2024.2434121
GitHub / repo	https://github.com/oxpig/Humatch
Service / docs	NR
Лицензия	BSD-3-Clause
Доступность / commercial-use	Code + web service

Задача и generated object

Used in these tasks: Humanization, Framework redesign

Part of these workflows: Framework humanization

Поле	Значение
Required input	Paired non-human VH/VL sequences
Conditioning	CDRs excluded by default; target human V-genes; paired-chain compatibility
Fixed	NR
Editable	NR
Generated	Humanized heavy/light variable-domain sequences and mutation set
Single mutations	Yes
Combinatorial mutations	Yes
Multi-objective	Humanness + pairing; not affinity/developability Pareto optimization

Architecture and operating principle

Author-reported (matrix)

Поле	Значение
Base architecture	NR
Model type	7 Humanization/framework redesign workflow
Representation	NR
Generation unit	Humanized heavy/light variable-domain sequences and mutation set
Conditioning	CDRs excluded by default; target human V-genes; paired-chain compatibility
Output	Humanized heavy/light variable-domain sequences and mutation set
Sampling	NR
Internal scorer	Human V-gene classifiers + pairing model
Structure generation	No structure/complex predictor

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

CDRs excluded by default; target human V-genes; paired-chain compatibility

Generation mechanism

Gene-specific CNN classifiers + iterative mutation selection and VH/VL pairing score

Sampling: NR

Training objectives

Losses: NR

Mathematical details: NR

Integrated evaluation

Human V-gene classifiers + pairing model

Architectural limitations

Humanness is not immunogenicity.
CDR preservation does not guarantee binding retention.
no prospective validation.

Uses these architecture components: Multi-objective optimization

Generator versus evaluation

Component	Role	Architecture class	Integrated or external	Required or optional	Influences generation or post hoc	Runtime included or excluded
Iterative humanization mutation workflow	primary_generator	Gene-specific CNN classifiers + iterative mutation selection and VH/VL pairing score	integrated	required	generation	included
Human V-gene classifiers + pairing model	integrated_scorer	NR	integrated	optional	post_hoc	unknown
Structure, binding retention, developability and immunogenicity assessment	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	2 retrospective evaluation on 177 humanized antibodies
Retrospective	No prospective wet-lab validation in primary paper
Prospective binding	NR
Prospective function	NR
Developability	NR
Tested design count	No prospective wet-lab validation in primary paper
Success criterion	Retrospective similarity to experimental humanizations; no measured affinity retention
Independent validation	NR
Summary	Retrospective similarity to experimental humanizations; no measured affinity retention

Limitations and failure modes

Humanness is not immunogenicity.
CDR preservation does not guarantee binding retention.
no prospective validation.

Deployment: Immediate retrospective evaluation

Reproducibility, license, commercial use

Поле	Значение
Repo	https://github.com/oxpig/Humatch
License	BSD-3-Clause
Commercial-use status	Code + web service

Alternative tools: HuDiff, HuAbDiffusion

Complementary: IgLM, AntiFold

Related: HuDiff, HuAbDiffusion

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

Humanization, Framework

Primary paper и полный список источников

Humatch: fast and accurate antibody humanization

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

Детальный source-grounded разбор

1. Что именно делает инструмент

[MATRIX]Вход: Paired non-human VH/VL sequences

[MATRIX]Generated object: Humanized heavy/light variable-domain sequences and mutation set

[MATRIX]Фиксируется: NR. Изменяется: NR.

[MATRIX]De novo или parent-based: Optimizes existing antibody.

2. Архитектура в одном абзаце

[MATRIX]Gene-specific CNN classifiers + iterative mutation selection and VH/VL pairing score

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: Humanized heavy/light variable-domain sequences and mutation set. Structure generation: No structure/complex predictor.

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): Retrospective similarity to experimental humanizations; no measured affinity retention

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): 2 retrospective evaluation on 177 humanized antibodies. Prospective binding: NR. Function: NR. Developability: NR.

Humanness is not immunogenicity.
CDR preservation does not guarantee binding retention.
no prospective validation.

15. Reproducibility

NRРепозиторий: https://github.com/oxpig/Humatch. Лицензия: BSD-3-Clause. Commit: NR, пока страница не имеет отдельного analysis YAML.

Модульная схема

Модуль	Класс модели	Вход	Выход	Обучается или frozen	Generator или evaluator
Iterative humanization mutation workflow	Gene-specific CNN classifiers + iterative mutation selection and VH/VL pairing score	NR	NR	NR	primary_generator
Human V-gene classifiers + pairing model	NR	NR	NR	NR	integrated_scorer
Structure, binding retention, developability and immunogenicity assessment	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
Iterative humanization mutation workflow	primary_generator	integrated	required	generation
Human V-gene classifiers + pairing model	integrated_scorer	integrated	optional	post_hoc
Structure, binding retention, developability and immunogenicity assessment	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
Retrospective similarity to experimental humanizations; no measured affinity retention	Gene-specific CNN classifiers + iterative mutation selection and VH/VL pairing score	2. retrospective benchmark	No prospective wet-lab validation in primary paper	NR	Retrospective similarity to experimental humanizations; no measured affinity retention	[MATRIX] landscape matrix 2026-08-13	Не читать как полный разбор primary paper.