DeepCryoRNA - deep learning-based RNA structure reconstruction from cryo-EM maps

Platform Requirements

• GNU/Linux x86_64 (tested on Ubuntu 16.04.7 LTS, Ubuntu 18.04.4 LTS, Ubuntu 20.04.6 LTS, Ubuntu 22.04.3 LTS)

Installation

It usually takes several minutes for the each installation step below.

1. Install Chimerax

Please download and install Chimerax Release 1.3 for Ubuntu 16, 18, 20 and Release 1.6 for Ubuntu 22 from https://www.rbvi.ucsf.edu/chimerax/older_releases.html.

Please check if Chimerax is installed:

chimerax --version

2. Install Anaconda with Python >= version 3.10

Please download and install Anaconda from https://www.anaconda.com/download#downloads.

3. Install TensorFlow 2

Please download and install Tensorflow following the website https://www.tensorflow.org/install/pip.

Please install TensorFlow with the pip package manager in Python from the newly installed Anaconda software.

Please follow the installation instructions for TensorFlow to install the NVIDIA software for GPU support. If not installed properly, only CPUs are used to predict atom classes by the neural network, which will slow down the prediction process.

4. Clone this repository on your local machine

git clone https://github.com/Vfold-RNA/DeepCryoRNA.git ${HOME}/DeepCryoRNA

5. Download the trained Unet model and put it in ${HOME}/DeepCryoRNA/src/

wget https://github.com/Vfold-RNA/DeepCryoRNA/releases/download/v1.0/DeepCryoRNA_Unet.hdf5 -P ${HOME}/DeepCryoRNA/src

6. Install the required Python packages, compile QRNAS ^[1] for energy minimization, and generate shared C++ library for efficient calculation of global sequence alignment scores

cd ${HOME}/DeepCryoRNA

Please change "Path_To_Anaconda" in the following command to the real path to the newly installed Anaconda:

bash ./install.sh Path_To_Anaconda

7. Set the environment variable DEEPCRYORNA_HOME

Add the following line to ${HOME}/.bashrc

export DEEPCRYORNA_HOME="${HOME}/DeepCryoRNA/src"

Open a new terminal and check the environment variable DEEPCRYORNA_HOME:

echo $DEEPCRYORNA_HOME

Usage of DeepCryoRNA

Run DeepCryoRNA for the example RNA 6UES

cd ${HOME}/DeepCryoRNA/Examples/example_6UES

Please change "Path_To_Anaconda" in the following command to the real path to the newly installed Anaconda:

Path_To_Anaconda/bin/python ${DEEPCRYORNA_HOME}/main.py -i ./input_6UES.txt > DeepCryoRNA_6UES.log

The input file "input_6UES.txt" includes the information for the RNA name, the cryo-EM map, RNA sequence, contour level, gpu and cpu. Please see the README.md in the Examples/ folder for further information.

The log file DeepCryoRNA_6UES.log stores the progress information.

Please see the README.md in the Examples/ folder for the information regarding the output files and folders.

For this example, it took about 8 minutes using a GPU and 10 CPUs.

It took about 10 to 15 minutes using only 10 CPUs.

Software References

[1] https://genesilico.pl/software/stand-alone/qrnas