Analyse data from the Amnis FlowSight to facilitate machine learning and 3rd party analysis.
Export features for each particle as .txt files using IDEAS software. Create a dictionary that links an "experiment name" to the exported .txt file. Create a folder called data/"experiment name"/ and export all particle images as tiff files to this directory for each experiment. Tip: export only the channels that you need to save time. This package uses Scanpy to perform dimentionality reduction and KNN based clustering, and tifffile to read the ome.tif files.
Use dataset.load() to read in all data into a pandas dataframe. then dataset.to_adata() to convert this dataframe into a workable AnnData object. Only non flourescent features are used as variables. These are transformed and scaled to work with PCA. Other features are added as observations to facilitate explorative plotting. These observations are only transformed. Use plot.PCA() to run pca and plot an overview. Use plot.cluster() to find clusters in the data based on a neighberhood graph and the leiden algorithm. Use plot.group_sample() to get sample images of particles from specified groups. Use this to find out where your particles of interest are, and how many groups ofthese that you have.
See "test/Test notebook.ipynb" for an example of all functions.