LucaOne: Generalized Biological Foundation Model with Unified Nucleic Acid and Protein Language.
- 2024/10/01: optimized embedding inference code:
src/get_embedding.py - 2024/08/01: add
checkpoint=17600000, location: checkpoint-step17600000 - 2024/07/24: feature: add
continue training when failure
Fig. 1 The workflow of LucaOne.
Fig. 2 The data and tasks for pre-training LucaOne, and T-SNE on four embedding models.
Fig. 3 Learning Central Dogma of Molecular Biology.
Fig. 4 Downstream task network with three input types and results comparison of 8 verification tasks.
sudo yum update
sudo yum install git-all
sudo apt-get update
sudo apt install git-all
wget https://repo.anaconda.com/archive/Anaconda3-2022.05-Linux-x86_64.sh
sh Anaconda3-2022.05-Linux-x86_64.sh
source ~/.bashrc
conda create -n lucaone python=3.9.13
conda activate lucaone
pip install -r requirements.txt -i https://pypi.tuna.tsinghua.edu.cn/simple
You can use the project: LucaOneApp Github or LucaOneApp FTP for embedding inference. For details, please refer to the README of the LucaOneApp project.
The project will download automatically LucaOne Trained-CheckPoint from FTP.
This project: LucaOneTasks Github or LucaOneTasks FTP is all the downstream tasks used in our paper(based on LucaOne's Embedding), and you can use this project to run other tasks, please refer to the README of this project.
Pretraining Dataset FTP: Dataset for LucaOne
Copy the dataset from http://47.93.21.181/lucaone/PreTrainingDataset/dataset/lucagplm into the directory: ./dataset/
The training dataset(dataset/lucagplm/v2.0/train/) whose file names start with '2023112418163521' are gene data(DNA + RNA), and those that start with '2023112314061479' are protein data.
The validation dataset(dataset/lucagplm/v2.0/dev/) whose file names start with '2023112418224620' are gene data(DNA + RNA), and those that start with '2023112314080544' are protein data.
The testing dataset(dataset/lucagplm/v2.0/test/) whose file names start with '2023112418231445' are gene data(DNA + RNA), and those that start with '2023112314083364' are protein data.
Notice
If you want to train individual nucleic acid or protein LucaOne(LucaOne-Gene or LucaOne-Prot), please separate the datasets as described above.
Training scripts are under the directory src/training, including 4 shell scripts:
run_multi_v2.0.sh: nucleic acid(DNA+RNA) and protein mixed training with 10 pre-training tasks.
run_multi_mask_v2.0.sh: nucleic acid(DNA+RNA) and protein mixed training with only 2 mask pre-training tasks.
run_multi_v2.0_gene.sh: individual nucleic acid training with 3 pre-training tasks.
run_multi_v2.0_prot.sh: individual protein training with 7 pre-training tasks.
run_multi_v2.0_continue.sh: continue training when failure.
FTP:
Pre-training data, code, and trained checkpoint of LucaOne, embedding inference code, downstream validation tasks data & code, and other materials are available: FTP.
Details:
The LucaOne's model code is available at: LucaOne Github or LucaOne.
The trained-checkpoint files are available at: TrainedCheckPoint.
LucaOne's representational inference code is available at: LucaOneApp Github or LucaOneApp.
The project of 8 downstream tasks is available at: LucaOneTasks Github or LucaOneTasks.
The pre-training dataset of LucaOne is opened at: PreTrainingDataset.
The datasets of downstream tasks are available at: DownstreamTasksDataset .
The trained models of downstream tasks are available at: DownstreamTasksTrainedModels .
Other supplementary materials are available at: Others .
Yong He, Zhaorong Li, Yongtao Shan, Pan Fang, Yanhong Wei, Yuan-Fei Pan, Mang Shi, Jiaying Yang, Yihao Chen, Yan Sun, Yuqi Liu
We have uploaded the model code, training scripts, and embedding inference scripts of LucaOne;
The mode code, training and evaluation scripts, datasets, and trained models for downstream tasks,
and additional supplementary materials to Zenodo (10.5281/zenodo.14977739).
However, due to the substantial size of the pretraining dataset of LucaOne, it has not been included on Zenodo.
Instead, it remains accessible via our publicly available FTP server (LucaOne Pretraining Dataset).
We are actively seeking an open FTP platform with sufficient storage capacity to host our pretraining dataset.
@article {LucaOne,
author = {Yong He and Pan Fang and Yongtao Shan and Yuanfei Pan and Yanhong Wei and Yichang Chen and Yihao Chen and Yi Liu and Zhenyu Zeng and Zhan Zhou and Feng Zhu and Edward C. Holmes and Jieping Ye and Jun Li and Yuelong Shu and Mang Shi and Zhaorong Li},
title = {LucaOne: Generalized Biological Foundation Model with Unified Nucleic Acid and Protein Language},
elocation-id = {2024.05.10.592927},
year = {2024},
doi = {10.1101/2024.05.10.592927},
publisher = {Cold Spring Harbor Laboratory},
URL = {https://www.biorxiv.org/content/early/2024/05/14/2024.05.10.592927 },
eprint = {https://www.biorxiv.org/content/early/2024/05/14/2024.05.10.592927.full.pdf },
journal = {bioRxiv}
}
Fig. 5 LucaTeam at the West Lake in Hangzhou.