cryoCheck

cryocheck is a deep learning-based cryo-EM micrograph quality assessment tool.

There are many kinds of contaminated cryo-EM images that can not contribute to the final reconstruction.

After taken your micrographs, either in format of mrc or png, cryocheck will automatically pick out good micrographs by excluding these types above.

Architecture

Installation

Make sure you have conda, CUDA(version >= 11.0), and cuDNN installed
Clone this repo:

git clone https://github.com/nzhou26/cryoCheck

Create conda environment

conda env create -f cryocheck_env.yml

Install dependencies from pip

# activate the environment just created
conda activate cryocheck

# install dependencies using pip
pip install -r requirement.txt

Download the latest model file. Edit MODEL_PATH in cryochekc_infer_png.py

Usage

Activate conda environment first

conda activate cryocheck

Pick out micrographs in mrc format:

cryocheck_infer_mrc.py /your/mrc/dir/

Pick out good mrc and tif so you can remove bad data

cleanup.py /your/good_mrc/dir/ /your/total_tif/dir/

Acknowledgement

All cryo-EM micrographs used in training are collected at Dr. Jun He's lab in Guangzhou Institute of Biomedicine and Health, Chinese Academy of Sciences

For any problems, please contact nzhou26@outlook.com