Python Template Matching Engine (PyTME)
Documentation | Installation | API
PyTME is a Python library for data-intensive n-dimensional template matching using CPUs and GPUs.
With its backend-agnostic design, the same code can be run on diverse hardware platforms using a best-of-breed approach. The underyling abstract backend specification allows for adding new backends to benefit from gains in performance and capabilities without modifying the library's core routines. The implementation of template matching scores is modular and provides developers with a flexible framework for rapid prototyping. Furthermore, pyTME supports a unique callback capability through analyzers, which allows for injection of custom code, enabling real-time processing and manipulation of results.
PyTME includes a graphical user interface that provides simplified mask creation, interactive filter exploration, result visualization, and manual refinement capabilities. This GUI serves as an accessible entry point to the library's core functionalities, allowing users to efficiently interact with and analyze their data.
Finally, pyTME offers specialized tools for cryogenic electron microscopy data, such as wedge masks, CTF correction, as well as means for handling structural data. Through dedicated integrations, the output of pyTME seamlessly integrates with commonly used cryogenic electron microscopy software such as RELION, Dynamo and IMOD.
Running into bugs or missing a feature? Help us improve the project by opening an issue.
Installation
We recommend installation using one of the following methods
| Method | Command |
|---|---|
| PyPi | pip install pytme |
| Source | pip install git+https://github.com/KosinskiLab/pyTME |
| Docker | docker build -t pytme -f docker/Dockerfile_GPU . |
You can find alternative installation methods in the documentation.
User Guide
Learn how to get started with
- Preprocessing: Picking the right mask and filter for template matching.
- Template matching: Find your template of interest.
- Postprocessing Analyze template matching results and downstream integrations.
How to Cite
If pyTME contributed significantly to your research, please cite the corresponding publication on SoftwareX.
@article{Maurer:2024aa,
author = {Maurer, Valentin J. and Siggel, Marc and Kosinski, Jan},
journal = {SoftwareX},
pages = {101636},
title = {PyTME (Python Template Matching Engine): A fast, flexible, and multi-purpose template matching library for cryogenic electron microscopy data},
volume = {25},
year = {2024}}