filament_tool_topaz

filament_tool_topaz is an application to process topaz machine learning picking jobs conducted on micrographs of flamentous molecules.

The filament fitting tool is advantageous as it eliminates false/dirty coordinate clusters, improving the efficiency of cryoEM 2D classification jobs as well as identifying equidistant neighbours to improve helical processing of molecules such as microtubules by exploiting their symmetry along the helix.

Set up the requirements

Download topaz here: https://github.com/tbepler/topaz

Use the package manager pip to install the requirements for filament_tool_topaz: pandas and scikit-learn

pip install scikit-learn
pip install numpy
pip install pandas

If you prefer to use anaconda, and have topaz installed in a conda environment, activate the conda environment and install the requirements using

conda activate [topaz_environment_name]
conda install [the packages above]

Set up the repository

git clone [this repository]
cd filament_tool_topaz

Instructions

To work on a newly generated topaz coordinate file, typically in the following format:
image_name x_coord y_coord score
Microtubules_00000 420 586 0.6130168
Microtubules_00000 881 475 -0.32318813
Microtubules_00000 514 359 -0.6220623
Microtubules_00000 541 107 -0.7319106
Microtubules_00000 578 382 -0.7987223
Microtubules_00000 489 529 -0.81050706
...

python src/topaz_to_filament.py [topaz_coordiante_file_PATH] [destination_folder_PATH]

This slices the topaz coordiante file into corresponding files for each micrograph, files of this type are referred to as "filament coordinate file" format.

To perform filament fitting on filament coordinate files, run the following command, which generates the fitted coordinate files using DBSCAN clustering and RANSAC regression, as well as png figures for each micrograph showing the results.

python src/filament_trace.py [destination_folder_PATH (ie. the filament coordinate files' location)]\
                             [destination_folder_PATH for fitted filament coordiantes]\
                             [spacing (euclidean distance between neighbouring fitted coordinates)]\
                             -t [topaz_particle_threshold]\
                             -eps [max distance between two coordinates to be considered neighbours for DBSCAN fitting]\
                             -min_samples [The number of samples in a neighbourhood for a point to qualify as a "core" [point]
                             --im [if set to 1, will save a png figure for each fitted example for visualization]
                             -box [box_size of the particles picked in pixels]

see example below, typically, eps depends on the resolution of the micrographs, but min_samples should be set to around 3-7

python src/filament_trace.py filament_mode_coordinates fitted_plots 10 -t -3.25 -eps 10 -min_samples 4 -im 1 -box 10

Results

Microtubules_00015.png

Pre-fitting:

Post-fitting: equidistant cooridnates along filaments for helical processing

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate.