LmPro-Nano: a method for nanobody optimization using pretrained language and energy prediction models
Nanobodies are an important therapeutic scaffold for multiple diseases and have been shown to neutralize severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). However, optimizing affinity for a chosen antigen remains slow and expensive. Here, we present LmPro-Nano, a sequence-driven workflow coupling a pretrained nanobody language model with a structure-based binding-energy predictor and molecular dynamics. We demonstrate LmPro-Nano on 3-2A2-4 and VHH72, two nanobodies against the SARS-CoV-2 receptor-binding domain. We find that LmPro-Nano improves antigen-binding-loop recovery by 4.37 points over a frequency reference, and for VHH72, sequence-only localization of the binding region raised the top-ten hit rate from two to seven tighter binders.
Authors
- Shuyuan Cao
- Feng Gao (ORCID: https://orcid.org/0000-0002-9563-3841)
- Bo Sun (ORCID: https://orcid.org/0009-0006-7067-1100)
Institutions
- Tianjin University (CN)
Publication Details
- Journal
- Genome biology
- Published
- 2026-11-21
- DOI
- https://doi.org/10.1186/s13059-026-04355-w
- Primary Topic
- Monoclonal and Polyclonal Antibodies Research
- Type
- article
- Field-Weighted Citation Impact
- 0.00