IgLM / AntiBERTa benchmarks

Antibody LM eval — paratope prediction, CDR generation, developability.

Composite score: 77.5

Rubric (1–5 per criterion)

rigor

4/5

coverage

4/5

maintenance

3/5

adoption

4/5

quality

4/5

accessibility

4/5

industry_relevance

4/5

Metadata

Stages

Hit IDDevelopmental Candidate

Modalities

biologic-mab

Task types

antibody-generationliability-prediction

License

MIT

First release

2022

Last updated

2024-08

Flags

none

Size & scope

sequences: 600000000
tasks: 6

Primary paper

Title

Generative language models for antibody design

Authors

Shuai RW, Ruffolo JA, Gray JJ

Year

2023

DOI / arXiv

10.1016/j.cels.2023.07.001

Citations

140

Hosted by (initiatives)

not hosted by any tracked initiative

Experts (primary authors / maintainers)

Jeffrey J. Gray
richard-shuai

Groups (host labs / companies / consortia)

Gray Lab (Johns Hopkins)

Related benchmarks

Observed Antibody Space (OAS)
SAbDab

Notes (honest caveats)

Moves toward true developability benchmarks.