Cryosparc-tools-scripts

This repository contains a set of scripts for processing CryoEM data using the CryoSPARC tools python library.

Contents

• crYOLO particle picking
  • crYOLO_particlepicker.py
  • crYOLO_trainedpicker.py
• cryodrgn
  • cryodrgn_trainer_downsampled.py
• cs2star
  • cs2star_2Dparticles.py

Scripts:

crYOLO particle picking

Particle picking scripts inspired from https://tools.cryosparc.com/examples/cryolo.html.

Installation

Create conda environment following the tutorial from the link above. But in short:

conda create -n cryolo -c conda-forge \
   python=3 numpy==1.18.5 \
   libtiff pyqt=5 wxPython=4.1.1 adwaita-icon-theme
conda activate cryolo
pip install -U pip
pip install nvidia-pyindex
pip install cryolo[c11] cryosparc-tools

crYOLO_particlepicker.py

crYOLO_particlepicker.py script enables you to perform crYOLO particle picking in CryoSPARC.

The script takes Curated Exposures and predicts particle locations. Particle picks can then be filtered by threshold in an Inspect Picks job.

The script takes the following command-line arguments:

• project - Name of the project to run the job in.
• workspace - Name of the workspace to run the job in.
• curate_exposures_job_id - ID of the job that curated the micrographs.
• box_size - Box size for particle picking (in Angstroms).
• model_path - Path to crYOLO model

Here is a sample command:

python crYOLO_particlepicker.py P1 W1 J3 110 path_to_model.h5

---

crYOLO_trainedpicker.py

crYOLO_trainedpicker.py script enables you to train crYOLO particle picking model and use it to pick particles in CryoSPARC.

The script takes in picked particles for training, usually from Select 2D job. Another importatnt input is the Exposure Sets job with split micrographs for training and testing. Particle picks can then be filtered by threshold in an Inspect Picks job.

Note: (1000/3000 micrographs split takes about 8 hours for training and prediction)

The script takes the following command-line arguments:

• project - Name of the project to run the job in.
• workspace - Name of the workspace to run the job in.
• training_particles_job_id - ID of job with picked particles for training (usually Select 2D job)
• exposure_sets_job_id - ID of exposure sets tool job, used to split micrographs
• box_size - Box size for particle picking (in Angstroms).
• model_path - Path to crYOLO model

Here is a sample command:

python crYOLO_trainedpicker.py P1 W1 J3 J5 110

---

cryodrgn

Script for running cryodrgn in cryosparc.

Installation

Before use, create conda environment for cryodrgn and make sure pytorch is working on your GPU.

conda create --name cryodrgn python=3.9
conda activate cryodrgn
pip install cryodrgn
pip install cryosparc-tools

cryodrgn_trainer_downsampled.py

cryodrgn_trainer_downsampled.py script enables you to train cryodrgn model and do basic analysis in CryoSPARC.

To prepare the inputs, run C1 Homogenous Refinement job and perform Downsample on selected particle set (usually box size of 128-256pix).

Note: For now, results can be found in job directory

The script takes the following command-line arguments:

• project - Name of the project to run the job in.
• workspace - Name of the workspace to run the job in.
• refinement_job_id - ID of the C1 Homogeneous Refinement job
• downsample_job_id - ID of the downsample particle job

Here is a sample command:

python cryodrgn_trainer_downsampled.py P1 W1 J10 J11

cs2star

Scripts using pyem (https://github.com/asarnow/pyem) in cryosparc.

TODO:

• input cryosparc job, run relion schema and output results back in cryosparc

installation

Create pyem conda environment before use:

conda create -n pyem
conda activate pyem
conda install numpy scipy matplotlib seaborn numba pandas natsort
conda install -c conda-forge pyfftw healpy pathos
git clone https://github.com/asarnow/pyem.git
cd pyem
pip install --no-dependencies -e .
pip install cryosparc-tools

cs2star_2Dparticles.py

cs2star_2Dparticles.py script enables you to directly prepare particles for Relion by just specifiying project, Select 2D job and path to relion projct folder.

Note: Script does not create a job in CryoSparc. It just sources the paths an folders

The script takes the following command-line arguments:

• project - Name of the project with 2D select job.
• select2D_job_id - ID of the Select 2D job.
• relion_project_path - Path to the RELION project

Here is a sample command:

python cs2star_2Dparticles.py P2 J12 path_to_relion_project