tribolium-clustering 0.2.5

A library for image ananalysis and data analysis of Tribolium embryos

Tribolium Clustering

This is a library that is specifically used for image analysis and data analysis of developing Tribolium embryos. It is optimised for stained nuclei of the embryos and 3D imaging. Most functions are wrappers for other libraries, mostly pyclesperanto, scikit-image, scikit-learn, UMAP and HDBSCAN.

At the moment some packages required need to be installed manually via conda. HDBSCAN needs to be installed via the command:

C:\Users\yourusername>conda install -c conda-forge hdbscan 

Similarly, a prerequisite for pyclesperanto_prototype: pyopencl needs to be installed with the command:

C:\Users\yourusername>conda install -c conda-forge pyopencl

Unfortunately the installation won't work without these.

The Image analysis is based almost entirely around functions from pyclesperanto which requires a powerful GPU with high memory capacities to work with high resolution files. If memory is insufficient you will probably encounter errors when using functions of this library and you need to make sure that you specify your powerful GPU in pyclesperanto with the command:

import pyclesperanto_prototype as cle

# "GTX" has to be replaced with the identifyer for your main GPU
cle.select_device("GTX")

The data analysis is based on scikit learn functions as well as other external data science algorithms that follow a similar convention. The basics of which can be found here. Since the functions were developed with a group of datasets modification of the functions could be required for the image analysis workflows but possibly also the data analysis workflows.