Installation and Download

System requirements

At our institute in Dortmund, crYOLO is running on the following operation systems:

  • Ubuntu 16.04.4 LTS
  • Ubuntu 18.04 LTS
  • CentOS 7

We don’t test it but it should run on Windows as well.

Moreover the following GPUs are used:

  • NVIDIA Titan V
  • NVIDIA GTX 1080
  • NVIDIA GTX 1080Ti
  • NVIDIA RTX 2080 TI
  • NVIDIA GV 100

As the GPU accelerated version of tensorflow does not support MacOS, crYOLO does not support it either.

crYOLO depends on CUDA Toolkit and the cuDNN library. These will be automatically installed during crYOLO installation.

Install crYOLO

Note

Please read the Complimentary Science Software License before using crYOLO. If you are interested in using crYOLO in a commercial context please contact stefan.raunser@mpi-dortmund.mpg.de

The following instructions assume that pip and anaconda or miniconda are available. In case you have a old cryolo environment installed, first remove the old one with:

conda env remove --name cryolo

With CUDA 10

The first step is to create a new virtual environment:

conda create -n cryolo -c conda-forge -c anaconda pyqt=5 python=3.7 cudatoolkit=10.0.130 cudnn=7.6.5 numpy==1.18.5 libtiff wxPython=4.1.1  adwaita-icon-theme

Activate the environment:

conda activate cryolo

In case you run crYOLO on a GPU run:

pip install 'cryolo[gpu]'

But if you want to run crYOLO on a CPU run:

pip install 'cryolo[cpu]'

With CUDA 11

Some graphic cards (e.g. RTX3090, A5000) need CUDA 11. Unfortunately, the official tensorflow 1.15x does not support CUDA 11. To get support for it, we need use a custom tensorflow version from nvidia. The following steps do explain how setup crYOLO with this custom nvidia.

The first step is to create a new virtual environment:

conda create -n cryolo -c conda-forge -c anaconda pyqt=5 python=3 numpy==1.18.5 libtiff wxPython=4.1.1  adwaita-icon-theme

Activate the environment:

conda activate cryolo

Next you need to installed the custom tensorflow version from nvidia:

pip install nvidia-pyindex
pip install nvidia-tensorflow[horovod]

To install crYOLO with CUDA 11 support you need to run:

pip install 'cryolo[c11]'

Warning

In case you run into glibc errors, you can find a solution in our troubleshooting section

Hint

You can also integrate crYOLO as Environment Module

That’s it!

You might want to check if everything is running as expected. Here is a reference example:

Reference example with TcdA1

Download the general models

We provide three general models. One for cryo-EM images which was trained on low-pass filtered images, another one for cryo-EM images but trained for images denoised by JANNI and one for negative stain images.

For cryo images (low-pass filtered)

Datasets:43 real, 10 simulated, 10 particle free datasets on various grids with contamination
Uploaded:27 May 2020
Download:ftp https
Config:Commands to create the config file can be found here.

For cryo images (neural network denoised with JANNI)

Datasets:43 real, 10 simulated, 10 particle free data sets on various grids with contamination
Uploaded:27 May 2020
Download:ftp https
Config:Commands to create the config file can be found here.

For negative stain images

Datasets:10 real data sets
Uploaded:26 February 2019
Download:ftp https
Config:Commands to create the config file can be found here.