Awesome Antibody Tools

List of databases and tools to discover, develop, and engineer antibodies. The goal is to create a comprhensive list of tools and databases for antibody-related predictions. With this effort we are also aiming to create a gold standard for benchmarking antibody-antigen predictions.

Datasets

This repository contains a manually curated dataset of protein-antigen pairs that are not likely to be used for training of the methods mentioned below

• https://adcdb.idrblab.net/

• Antibody Registry

• Observed Antibody Space (OAS)

• https://opig.stats.ox.ac.uk/webapps/plabdab, github: https://github.com/brennanaba/PLAbDab

• http://research.naturalantibody.com/paddocumentation

Tools

AI-based antibody-antigen prediction

• AbAgIntPre: - http://www.zzdlab.com/AbAgIntPre/ https://www.frontiersin.org/journals/immunology/articles/10.3389/fimmu.2022.1053617/full https://github.com/emersON106/AbAgIntPre

• ImaPEp: - https://www.mdpi.com/1422-0067/25/10/5434 https://github.com/3BioCompBio/ImaPEp

• GeoFlow: not open-source but available for non-commercial applications

• Geometric Paratope Epitope Prediction: https://github.com/Marco-Peg/GEP

• Antipasti: binding affinity prediction https://github.com/kevinmicha/ANTIPASTI

• AbSciBio/unlocking-de-novo-antibody-design
  Generative AI for de novo design of antibody sequences.
  Repository includes datasets and design outputs for novel antibody candidates.

• AIforGreatGood/biotransfer
  Machine learning–driven framework for designing high-affinity scFv libraries.
  Uses Bayesian language models for in silico candidate optimization.

• luost26/diffab
  Diffusion-based generative model for antigen-specific antibody design and optimization.
  Developed and presented at NeurIPS 2022. Benchmark: https://github.com/AstraZeneca/DiffAbXL

• pengzhangzhi/ab_opt
  Generative Diffusion Models for Antibody Design, Docking, and Optimization.
  Provides an end-to-end pipeline that integrates design with docking and experimental validation data.

• IgGM (TencentAI4S/IgGM)
  A generative model for functional antibody and nanobody design.
  Can design overall structures as well as specific CDR loops based on antigen and epitope inputs.

• LizeRaes/ai-drug-discovery
  A demo application using LangChain4j for AI-backed drug discovery research.
  Includes steps to derive new candidate antibodies from antigen sequence and known antibody features.

• GeoAB: https://github.com/EDAPINENUT/GeoAB

Docking-based antibody-antigen prediction

• GeoFlow - not open-source but available for non-commercial applications

• AbAdapt - predict most likely pose from AA sequences of antigen + antibody https://sysimm.org/abadapt/workflow

General PPI prediction tools that can be applied here

• Boltz2 https://github.com/jwohlwend/boltz?tab=readme-ov-file

• DiffDockPP - AI-driven MD for PPI

• AlphaFold

• HADDOCK

Antibody optimization tools

improve affinity

• AttABseq "Unsupervised evolution of protein and antibody complexes with a structure-informed language model". https://github.com/varun-shanker/structural-evolution https://www.science.org/stoken/author-tokens/ST-1968/full

• oxpig/AntiFold
  Antibody-specific inverse folding model for structure-based design.
  Fine-tuned from ESM-IF1 to predict sequences that fold into a desired antibody structure with optimized binding properties.

• MEAN: https://github.com/THUNLP-MT/MEAN

improve developelability

• gv20-therapeutics/antibody-in-pytorch
  PyTorch implementations of machine learning models for antibody sequence analysis.
  Includes models for affinity, developability, and general sequence evaluation.

• Mohammad-Vahed/AI-BioDiscover
  An AI/ML tool for antibody design and molecular property prediction.
  A demo app that walks through candidate discovery and property evaluation.

humanization

Other Antibody Tools

EvolvePro - https://github.com/mat10d/EvolvePro https://www.science.org/doi/10.1126/science.adr6006 https://github.com/idmjky/EvolvePro It requires experimental measure for optimization and training

Antibodies language

• AbLEF antibody language https://github.com/Merck/AbLEF https://academic.oup.com/bioinformatics/article/40/5/btae268/7646845

• Graylab/IgLM
  Generative language model for antibody design.
  Supports sequence infilling, full sequence generation, and sequence evaluation (e.g. redesigning CDR loops).

• oxpig/AbLang2
  An antibody-specific language model for predicting non‑germline (mutated) residues.
  Useful for guiding antibody humanization and optimization.

Structure prediction tools

• Graylab/IgFold
  State-of-the-art antibody structure prediction from sequence alone.
  Leverages pretrained language models (e.g. AntiBERTy) to rapidly predict antibody 3D structures.

• RoseTTAFold
  Deep learning-based protein (and antibody) structure prediction tool.
  Although not antibody‑specific, it is widely used in antibody modeling and design.

• Rosetta@home & RosettaCommons Tools
  A suite of computational tools for protein structure prediction, docking, and design.
  Accessible via public servers and repositories; note that many parts of Rosetta are licensed for academic use.

• HADDOCK-antibody-antigen
  Protocol for antibody–antigen docking using HADDOCK.
  Supports modeling of antibody–antigen complexes.

AB funnel pipeline

antigen -> antigen structure + seq -> predicted AB or starting AB -> AB structure prediction -> improve affinity -> improve developelability -> docking

• Platforma
  • Bioinformatics platform for antibody discovery from sequencing data. Covers design → optimization → lead selection in one environment. Extensible via an open-source SDK ), with reusable "blocks."

Contribute

To contribute open a new issue with your suggestions.