Artiatomi – Installation instructions

System requirements for CLI / command line applications:

For Windows 10 or newer:

  • A Nvidia GPU with compute capability 6.1 (Pascal) or better
  • CUDA 11.4 or newer
  • MPICH, can be downloaded here
  • For aiDoseBooster and aiDoseBoosterTraining: CUDNN from Nvidia or download from here: cudnn.zip
  • Visual C++ runtime, can be downloaded here

For Linux / Ubuntu 20.04 or newer:

  • A Nvidia GPU with compute capability 6.1 (Pascal) or better
  • CUDA 11.1 or newer
  • OpenMPI installed (sudo apt-get install openmpi-bin)
  • libtbb2 installed (sudo apt-get install libtbb2)
  • For aiDoseBooster and aiDoseBoosterTraining: CUDNN from Nvidia or download from here: cudnn.tar.gz

System requirements for GUI applications:

For Windows 10 or newer:

  • For aiCtfDetectorGui only: CUDA 11.4 or newer
  • Visual C++ runtime, can be downloaded here
  • (Qt 6.2, the windows package provides all necessary DLL)

For Linux / Ubuntu 20.04 or newer:

  • Qt 6.2, a set of binaries compiled by us can be downloaded here: Qt624.tar.xz
  • libtbb2 installed (sudo apt-get install libtbb2)
  • For aiCtfDetectorGui only: CUDA 11.1 or newer

For MAC (aiClicker only)

  • coming soon!

Installation – Windows

Download the Artiatomi package (Command line and/or GUI package) and extract it to any folder. If one uses the CLI-applications on Windows, it can be useful to set the PATH and ARTIATOMI environment variable to the extracted folder.

Installation – Linux/Ubuntu 20.04

Download the Artiatomi package (Command line and/or GUI package) and extract it to any folder. In case that the GUI applications are to be installed and no other Qt6.2 installation is available, download our compiled binaries (here: Qt624.tar.xz ) and extract them into the subfolder „lib“.
E.g.:
your_folder_to/artiatomi -> contains all artiatomi binaries
your_folder_to/artiatomi/lib -> contains libraries artiatomi depends on
your_folder_to/artiatomi/lib/Qt6 -> contains Qt6.2 libraries and folders

For the GUI package: In the folder with the artiatomi binaries one can find two desktop files: aiClicker.desktop and aiCtfDetectorGui.desktop. Open them in a text editor and adjust the paths accordingly to the chosen artiatomi folder. Then copy the two file to ~/.local/share/applications.

In ~/.bashrc add the three lines:

export ARTIATOMI_PATH=your_folder_to/artiatomi/
export LD_LIBRARY_PATH=your_folder_to/artiatomi/lib:your_folder_to/artiatomi/lib/Qt6/lib:$LD_LIBRARY_PATH
export PATH=your_folder_to/artiatomi/:$PATH

The following step is necessary for starting the GUI applications without a terminal:
As ’sudo‘ in folder /etc/ld.so.conf.d/ add an empty file ‚artiatomi.conf‘ and write these two lines into it:

/your_folder_to/artiatomi/lib
/your_folder_to/artiatomi/lib/Qt6/lib

Log out and log in again or reboot the PC.
Now, when double-clicking on aiClicker or aiCtfDetectorGui, the application should start.

Setting up artiatomi.settings

When importing dose-fractionation movie stacks into Artiatomi using a SerialEM-MDOC-file, not all necessary data is directly available. As most this information is constant over many datasets, Artiatomi allows to set this settings in a global artiatomi.settings file.

Settings up this file is easy when using aiToolbox. For each microscope you want to automatically import tilt-series from, you need a sample MDOC-file. The MDOC file contains a microscope specific identifier (usually on line 7 starting with "[T = SerialEM:") that is used to identify a specific setting.

You will setup the following values:
– The CS value of the objective lens in mm
– The acceleration voltage in kV
– The rotation in degrees to add to the image rotation indicated in the MDOC-file to match with Artiatomi convention
– Indication if the raw dose-fractionation images need to be flipped on the Y-axis
– Magnification anisotropy values
– Indication if the target or the measured defocus value in the MDOC-file are more precise. Depending on this choice, Artiatomi takes the indicated value as a first estimate for defocus
– Defocus multiplicative factor to apply to the value in the MDOC-file, usually -1000 to convert the value in µm to nm as used in Artiatomi, and negative to have positive values for underfocus
– Indication if the acceleration voltage is read from the MDOC-file or from the settings file, might be useful in case that a microscope is used with varying voltages.

For example, for a CS value of 2.7, additional rotation angle of -175 degrees and a magnification anisotropy of 1.016 and 42 degrees (all other values remain at default), one uses aiToolbox on a shell:

aiToolbox settings add --name path/to/the/mdoc.file --addRot -175 --cs 2.7 --magAniso 1.016 --magAnisoAngle 42

On the first call, the file artiatomi.settings will be created in the path defined in the environment variable ARTIATOMI_PATH. On every other call, a settings section will be added to the file if a new identifier is found in the MDOC-file.

As a final note, we emphasize that it is important to have the correct image flip and rotation values — otherwise the handedness is wrong and also 3D-CTF correction might apply the opposite defocus gradient which would result in non-satisfying results. As a general hint: The images should be displayed in aiClicker with the same orientation as in SerialEM, IMOD or Digital Micrograph. If they appear mirrored, one has to activate the flip-on-Y-axis option. The rotation angle in Artiatomi is usually relatively small for most settings: If the tilt-axis in SerialEM or IMOD is close to the X-axis of the image, the final rotation in Artiatomi is likely in the range [-20°..20°]. If on the other hand the tilt-axis in SerialEM or IMOD is closer to the Y-axis, one has to try out if the final rotation in Artiatomi is -90° or +90°. This is important as the additional rotation by 180° again swaps the handedness and the CTF-gradient direction. We had this issue initially with the EMPIAR-10499 dataset.