This is a PyTorch implementation of our paper:
Lei Sun*, Kui Xu*, Wenze Huang*, Yucheng T. Yang, Lei Tang, Tuanlin Xiong, Qiangfeng Cliff Zhang
*: indicates equal contribution.
bioRxiv preprint: (https://www.biorxiv.org/content/10.1101/2020.05.05.078774v1)
- Python 3.6
- PyTorch 1.0.0, with NVIDIA CUDA Support
- pip
Clone repository:
git clone https://github.com/kuixu/PrismNet.gitInstall packages:
cd PrismNet
python setup.py installScripts and pipeline are in preparing, currently, we provide a sample data in HDF5 format in data folder.
data
├── TIA1_Hela.h5
to train one single protein model from scratch, run
exp/EXP_NAME/train.sh TIA1_Hela
where you replace TIA1_Hela with the name of the data file you want to use, you replace EXP_NAME with a specific name of this experiment. Hyper-parameters could be tuned in exp/prismnet/train.sh. For available training options, please take a look at tools/train.py.
You can monitor on http://localhost:6006 the training process using tensorboard:
tensorboard --logdir exp/EXP_NAME/out
For evaluation of the models, we provide the script eval.sh. You can run it using
exp/EXP_NAME/eval.sh TIA1_Hela
Scripts of the analysis on integrative motifs, riboSNitch and structurally variable sites are orgnizing, which could be glanced at here.
git clone https://github.com/huangwenze/PrismNet_analysis
This project is free to use for non-commercial purposes - see the LICENSE file for details.
@article {Sun2020.05.05.078774,
title = {Predicting dynamic cellular protein-RNA interactions using deep learning and in vivo RNA structure},
author = {Sun, Lei and Xu, Kui and Huang, Wenze and Yang, Yucheng T. and Tang, Lei and Xiong, Tuanlin and Zhang, Qiangfeng Cliff},
year = {2020},
eprint = {https://www.biorxiv.org/content/early/2020/05/07/2020.05.05.078774.full.pdf},
journal = {bioRxiv}
}