angelolab / ark-analysis

Integrated pipeline for multiplexed image analysis
https://ark-analysis.readthedocs.io/en/latest/
MIT License
73 stars 26 forks source link

Make 99.9 normalization optional on function cell_som_clustering.train_cell_som #1000

Closed JLrumberger closed 1 year ago

JLrumberger commented 1 year ago

Is your feature request related to a problem? Please describe. I want to use the templates/generic_cell_clustering.ipynb on data from the cell classification model. @ngreenwald mentioned that I should run the SOM clustering without 99.9 percentile normalization on this data, but the function train_cell_som and the downstream class CellSOMCluster do not allow me to turn off normalization.

Describe the solution you'd like Add a flag to cluster_helpers.CellSOMCluster to the init called normalize with default True and add control flow to the init to make self.train_data = self.normalize_data(self.train_data) optional.

Add a flag to cell_som_clustering.train_cell_som called normalize with default True and provide this argument to cluster_helpers.CellSOMCluster.

Describe alternatives you've considered None

Additional context None

srivarra commented 1 year ago

congrats on issue / pr number 1000! Lorenz! 🥳