AstraZeneca/DiffAbXL

The official implementation of DiffAbXL benchmarked in the paper "Exploring Log-Likelihood Scores for Ranking Antibody Sequence Designs", formerly titled "Benchmarking Generative Models for Antibody Design".

39
/ 100
Emerging

Implements diffusion-based generative models for antibody sequence design with two operational modes—de novo generation and structure-guided design—evaluated via log-likelihood scoring for ranking sequences by predicted binding affinity. The framework benchmarks multiple model architectures (diffusion, LLM, and graph-based) across five experimental datasets with standardized interfaces, enabling direct comparison of ranking performance through Spearman correlation metrics. Built on PyTorch with configurable training pipelines and provides modular integration points for external models via standardized benchmark interfaces.

No commits in the last 6 months.

Stale 6m No Package No Dependents
Maintenance 2 / 25
Adoption 9 / 25
Maturity 16 / 25
Community 12 / 25

How are scores calculated?

Stars

90

Forks

9

Language

Python

License

Apache-2.0

Last pushed

Jun 11, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/generative-ai/AstraZeneca/DiffAbXL"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.