LucaOne
A generalized biological foundation model that unifies nucleic acid and protein language modeling in a single architecture, usable for sequence embedding, masked pretraining, and sequence- or token-level classification and regression tasks.
LucaOne, also called LucaGPLM, is a generalized biological foundation model that unifies nucleic acid and protein language into a single architecture, described in a paper published in Nature Machine Intelligence. It provides task-specific heads for embedding sequences, pretraining and sequence recovery through masked language modeling, and sequence-level or token-level classification and regression tasks such as protein family or solubility prediction.
It offers native Hugging Face support through standard auto-model, auto-tokenizer, and auto-config classes, and supports further training using sequence-only inputs or by injecting biological knowledge following the LucaOne framework, including fine-tuning for classification or regression.
It is developed by a research group at Alibaba Cloud, is written in Python, and is released under the Apache-2.0 license, with trained checkpoints available on Hugging Face.