GPU-accelerated template matching for cryo-electron tomography, originally developed in PyTom, as a standalone Python package that is run from the command line.
This software is developed by Marten L. Chaillet (@McHaillet) and Sander Roet (@sroet) in the group of Friedrich Förster at Utrecht University.
miniconda3
nvidia-cuda-toolkit
There are 2 options for creating a conda environment. We recommend option (1) which will later allow cupy to build against a system installed cuda-toolkit. Compared to option (2) this can give an almost two-fold speedup:
(recommended) Create a new python 3 environment:
conda create -n pytom_tm python=3
Create a new environment with a prebuild cupy version and complete CUDA-toolkit. This is more reliable, but takes more disk space and has less optimal performance.
conda create -n pytom_tm -c conda-forge python=3 cupy cuda-version=11.8
Once the environment is created, activate it:
conda activate pytom_tm
Then install the code with pip
(building cupy can take a while!):
python -m pip install pytom-match-pick[plotting]
The installation above also adds the optional dependencies [matplotlib, seaborn]
which are required to run
pytom_estimate_roc.py
. They are not essential to the core template matching functionality, so for some systems
(such as certain cluster environments) it might be desirable to skip them. In that case remove [plotting]
from the pip install command:
python -m pip install pytom-match-pick
Having issues running the software? If cupy is not correctly installed,
python -c "import pytom_tm"
can show a cupy warning. If this is the case, this probably means cupy is not correctly installed. Alternatively, cupy can sometimes be installed without issue but not detect CUDA correctly. In that case, the following should raise some errors:
python -c "import cupy as cp; a = cp.zeros((100,100))"
To solve cupy installation issues, please check the cupy docs. It might be solved by installing a specific build compatible with the installed cuda toolkit.
The following scripts are available to run with --help
to see parameters:
pytom_create_template.py --help
pytom_create_mask.py --help
pytom_match_template.py --help
pytom_extract_candidate.py --help
pytom_estimate_roc.py --help
pytom_merge_stars.py --help
Detailed usage instructions and a tutorial are available on our site.
Please use our github discussions for:
If you want the most up-to-date version of the code you can get install it from this repository via:
git clone https://github.com/SBC-Utrecht/pytom-match-pick.git
cd pytom-match-pick
python -m pip install '.[all]'
if you don't want the optional plotting dependencies use the following install command instead:
python -m pip install '.[dev]'
For development, please also install pre-commit to check and autostyle the code before you make PRs:
pre-commit install
This uses Ruff to check and format whenever you make commits.
If you update anything in the (documentation) docs/
folder make sure to test build the website locally:
mkdocs serve
With the developer install also comes the ability to run the unittests, from the git repository run:
cd tests
python -m unittest discover
Contributions to the project are very welcome! Feel free to make a pull request or suggest an implementation in the issues. For PR's we will gladly give you feedback on how to integrate the code.
Chaillet, M. L., van der Schot, G., Gubins, I., Roet, S., Veltkamp, R. C., & Förster, F. (2023). Extensive angular sampling enables the sensitive localization of macromolecules in electron tomograms. International Journal of Molecular Sciences, 24(17), 13375. https://doi.org/10.3390/ijms241713375