# LucaOne APP

Inference and embedding code for LucaOne, a foundation model for nucleic-acid and protein sequences, with Hugging Face support for fine-tuning on sequence- and residue-level biological tasks.

LucaOneApp provides the code to run embedding inference with LucaOne, a biological foundation model covering DNA, RNA and protein sequences. Model checkpoints, identified by a step number and version such as LucaOne, LucaOne-Gene (trained only on DNA and RNA) or LucaOne-Prot (trained only on protein), are downloaded automatically based on the requested version and step.

The project also exposes a Hugging Face-native interface, with AutoModel, AutoModelForMaskedLM, AutoModelForSequenceClassification and AutoModelForTokenClassification classes built on one shared architecture, so the model can be fine-tuned for sequence-level tasks such as protein family or solubility prediction, or residue-level tasks such as secondary structure and binding-site prediction, using the standard transformers API.

It is written in Python and released under the Apache-2.0 license.

Website: https://github.com/LucaOne/LucaOneApp