Skip to content

qnano/simflux

Repository files navigation

SIMFLUX Readme

SIMFLUX is implemented using C++/CUDA 10.1 and Python, running on Windows 10. A version with the binary already build can be found here

Installing python

If the binary is already there, you only need to install python and some python packages. It is recommended to use anaconda 3 or miniconda. Make sure to get python >= 3.6 on 64 bit: https://repo.anaconda.com/archive/Anaconda3-2019.07-Windows-x86_64.exe

conda create -n simflux_env python
conda activate simflux_env
conda install tqdm matplotlib scikit-image pyyaml scikit-image scikit-learn h5py
conda install -c conda-forge tifffile

Building C++ / CUDA library

(The order is important here)

  • Install Visual studio 2017 or 2019 Community Edition: https://visualstudio.microsoft.com/downloads/

  • Open visual studio. Menu "Tools / Get Tools and Features"

  • Install the Visual studio 2015 build toolset. We're using the Visual Studio 2015 build tools because we ran into cuda compiler issues at 2017.

  • Install CUDA Toolkit 10.1 https://developer.nvidia.com/cuda-downloads If you install this before visual studio 2017 and the 2015 toolset, you need to reinstall it to make sure the CUDA BuildCustomizations are installed.

  • Download the external libraries below and extract in the root. There should be a folder "external" in the repository root. https://drive.google.com/open?id=1E7f7xqNLwqtj4_-2bjJ56T6kbzsdCpDQ

  • Open the SMLM.sln solution file. Don't upgrade the projects.

Testing

python/simflux_example.py can be used as a basic test and example of how to run the code. It will automatically download some sample data of around 200MB.

cd python
python simflux_example.py

The results will be in a folder named "results/object_wlc_15", with various plots and localization files.

  • simflux.hdf5 - SIMFLUX localizations
  • g2d.hdf5 - Conventional SMLM localizations

Linked localizations (Localizations in subsequent frames merged into one more accurate spot, for FRC calculation):

  • simflux-linked.hdf5 - SIMFLUX localizations
  • g2d-linked.hdf5 - Conventional SMLM localizations

These HDF5 files are compatible with the Picasso render application for viewing:

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published