Efficient biological design via genomic in-context learning
- 1Stanford University
Abstract
In-context learning (ICL), the ability to perform new tasks from a few examples supplied in a prompt, is a defining emergent capability of large language models. Here, we show that ICL emerges in the genome language model Evo 2 and enables biological design. With as few as four examples in the prompt and no additional training, Evo 2 generates realistic metal-binding proteins, de novo antibodies, protein-RNA systems, and entire bacteriophage genomes. For {Phi}X174-like bacteriophage design, ICL surpasses previously fine-tuned models on computational measures of design quality and diversity. Beyond {Phi}X174, conditioning the model with different phage prompts enables us to efficiently design MS2 and M13 genomes in silico without fine-tuning. Iteratively updating the prompt with newly generated phage genomes, which we refer to as genomic chain-of-thought, further improves the diversity of designs while maintaining quality metrics. Paired with an inference engine that accelerates generation up to 35-fold and extends practicable context to a megabase, these results establish ICL as a versatile and efficient paradigm for biological sequence design.
Related
- Scaling a context-aware genomic model for condition-specific edit design in plants
- Machine learning reveals sequence and genomic context features underlying Alu-specific effects on genome folding
- A Context-Aware Framework for Biological Health Index Harmonization and Machine Learning-Aided Health Forecasting of Waterbodies
- A Context-Aware 14-3-3 Binding Predictor Enabled by Data Augmentation, Multi-Scale Biological Features, and Protein Language Model Embeddings
- Locus coeruleus noradrenergic neurons have diverse activity patterns across species and biological context
BibTeX
@misc{brixi2026efficient,
title = {{Efficient biological design via genomic in-context learning}},
author = {G. Brixi and Y. A. Hao and D. Nori and A. Merchant and T. Widatalla and S. King and K. Shalvarjian and D. Park and B. Hie},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.28.762643},
url = {https://www.biorxiv.org/content/10.64898/2026.10.28.762643v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Efficient biological design via genomic in-context learning
AU - G. Brixi
AU - Y. A. Hao
AU - D. Nori
AU - A. Merchant
AU - T. Widatalla
AU - S. King
AU - K. Shalvarjian
AU - D. Park
AU - B. Hie
PY - 2026
DA - 2026/10/28
PB - bioRxiv
DO - 10.64898/2026.10.28.762643
UR - https://www.biorxiv.org/content/10.64898/2026.10.28.762643v1
AB - In-context learning (ICL), the ability to perform new tasks from a few examples supplied in a prompt, is a defining emergent capability of large language models. Here, we show that ICL emerges in the genome language model Evo 2 and enables biological design. With as few as four examples in the prompt and no additional training, Evo 2 generates realistic metal-binding proteins, de novo antibodies, protein-RNA systems, and entire bacteriophage genomes. For {Phi}X174-like bacteriophage design, ICL surpasses previously fine-tuned models on computational measures of design quality and diversity. Beyond {Phi}X174, conditioning the model with different phage prompts enables us to efficiently design MS2 and M13 genomes in silico without fine-tuning. Iteratively updating the prompt with newly generated phage genomes, which we refer to as genomic chain-of-thought, further improves the diversity of designs while maintaining quality metrics. Paired with an inference engine that accelerates generation up to 35-fold and extends practicable context to a megabase, these results establish ICL as a versatile and efficient paradigm for biological sequence design.
ER -