Spatial-LDA viz and preprocessing notebook

[bcollica]

What is the purpose of this PR?

This PR creates the notebook and associated functions/visualizations for preprocessing data for spatial-LDA.

How did you implement your changes

The LDA_Preprocessing notebook is similar to the other example notebooks in the sense that it walks the user through the steps of loading, formatting, and visualizing cell data for use in spatial-LDA analysis. processing.py is the main module with the formatting and processing functions. Some visualization functions were added to analysis/visualize.py, and a few random helper functions were added to utils/spatial_lda_utils.py.

The notebook uses an example dataset with two FOVs to show how the functions work. This dataset is included in example_dataset/spatial_lda_input_data. There are also directories with examples of the different visualizations as well as the processed output data which will get passed to the training/inference functions in the next step.

Remaining issues

A few points here:

1. The example dataset is kinda wonky because the centroids are pulled from a different cell table, and having only two FOVs doesn't really produce anything super interesting anyway. This will be resolved when an official example dataset has been decided on.
2. The bootstrapping in topic_eda() can be a bit slow, especially for larger sets of FOVs and also larger topic numbers. One thought is to add support for parallel processing either across topics or across bootstrap iterations.
3. All visualizations use default parameters for color and style, but this can be adjusted depending on preferences.
4. There is currently a function for writing/saving the different data components, but not a function for reading/loading them which will be useful in the training/inference notebook but also for other potential exploratory analysis.
5. The notebook contains a “table of contents” with internal links to the different sections, but these might not actually function properly depending on how the user is viewing the notebook (i.e. in a browser, in an IDE, etc.).

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[ngreenwald]

Looking good. You added a ton of files in this PR. I think it makes sense to have the input data, but I don't think we need to check in the outputs, especially the plots that are generated.

[alex-l-kong]

Just a few minor things to add.

[ngreenwald]

Looks good!