Software for spatial tissue profiling by imaging-free molecular tomography.

Resources
Installation
Use

Tutorials. Documentation and tutorials explaining the step-by-step usage of Tomographer can be found here: http://tomographer.info/
Example Notebooks. Step-by-step notebooks to be found here: https://github.com/lamanno-epfl/tomographer/blob/master/tutorials/
Tomographer Data Viewer. Tomographer data-set browser of mouse and lizard brains: https://strpseq-viewer.epfl.ch/
Original Article. https://www.nature.com/articles/s41587-021-00879-7

The installation of the tomographer package and all requirements is achieved in the following steps.

If you don't have conda, you might find it helpful to install Miniconda before beginning:

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh

First, configure your environment containing some dependencies:

conda create -n tomographer-env python=3.7 numpy=1.19.0 scipy=1.2.0 pandas=0.24.1 scikit-learn=0.20.2 scikit-image=0.14.2 matplotlib=3.0.2 --channel bioconda --channel conda-forge
conda activate tomographer-env

Second, install required libraries

pip install PyWavelets GPy GPyopt
git clone https://github.com/jmetzen/gp_extras.git
cd gp_extras
[sudo] python setup.py install

For some tutorial notebooks you will also need BrainMap:

git clone https://github.com/linnarsson-lab/brainmap
cd brainmap
[sudo] python setup.py install

Third, clone tomographer locally

git clone https://github.com/lamanno-epfl/tomographer.git

Lastly, install in development mode

pip install -e ./tomographer

To update you can simply pull (you don't need to reinstall):

cd tomographer
git pull

Please refer to http://tomographer.info/ for a detailed tutorial

The package requires 3 input files and 2 output directories.

• Config File : This is an .hdf5 formatted file which contains the following keys and values

  • angles _ names : array of angle names like ['angle9', 'angle117']
  • angles _ values : array of angle values in radians like [3.65, 5.096]
  • first _ points : array specifying the starting point to begin considering values within projection for each angle
  • mask _ g : 2D image array of mask (binary)
  • mask _ thrs : threshold used to create mask from reference image
  • masked _ formulation : Boolean indicating if design matrix should be recalculated using the masked formulation
  • proj _ len : array indicating the lengths of projections for each angle
  • reference _ mask : image array of reference image
  • symmetry : Boolean indicating if design matrix should be recalculated assuming axis of symmetry in tissue
  • widths : array indicating the estimated widths of the secondary slices in each angle
  • A : Specified design matrix. Note that this matrix is recalculated if the symmetry value is True or if the masked formulation value is True

• Input File : This is an .hdf5 formatted file which contains a dictionary of dictionaries. It can be created with the ReconstructionConfig Class

  • The key genes is further queried by
    • The gene name, which itself contains the keys containing
      • The angle names that match those in the config file
        • Values inside input_file['genes'][gene_name][angle_name] correspond to the projection values

• Gene File: Is a .txt formatted file which contains the individual gene names separated by lines (\n). These are queried by the tomographer one by one

• Output Reconstruction File : This is an .hdf5 formatted file which contains all the reconstructed genes
• Output Alpha-Beta File : This is an .hdf5 formatted file which contains all the selected alpha and beta values that were selected for a given reconstruction. This may be useful for filtering out poorly reconstructed genes.

From tomographer, one can run the following command:

python3 tomography/tomorun.py -c path_to_config.hdf5 -i path_to_inputs.hdf5 -g /list_of_genes.txt -o /path_to_output.hdf5 -a /path_to_/alpha_beta_output.hdf5