stimwrap 0.3.0

Quick start

We provide an example notebook under notebooks/openst_example.ipynb, and also a Google Colab notebook

These will walk you through downloading Open-ST data, running STIM in a Python notebook (via stimwrap bindings), and seamlessly running some downstream analyses.

We provide another notebook under notebooks/visium_example.ipynb, and also a 2nd Google Colab notebook showcasing the 10x Visium adult mouse brain dataset showcased in all our tutorials.

Installation

To install the stimwrap try:

pip install stimwrap

or if you do not have sudo rights:

pip install --user stimwrap

Check if the library is successfully installed:

python -c import stimwrap as sw

If installation fails due to conflicting dependencies, create a dedicated environment with python>=3.9 and try again to install stimwrap as above.

Usage

The following assumes that the file container.n5 contains the datasets and their attributes as created by STIM:

import stimwrap as sw
pucks = sw.Container('/path/to/container.n5')

Print the names of the datasets:

print(pucks.get_dataset_names())

Focus on a specific puck and extract the relevant information:

puck_name = pucks.get_dataset_names()[0]
puck = pucks.get_dataset(puck_name)

Get the puck locations either directly from the puck:

locations = puck['locations']

or fetch them from the container:

locations = pucks.get_dataset(puck_name)['locations']

The examples above assume that the dataset is N5-backed. For AnnData-backed datasets, the key for the puck locations might be:

locations = pucks.get_dataset(puck_name)['spatial']

which will try to access the obsm/spatial variable from the dataset. Alternatively, we recommend using the official AnnData package for handling these files.

hpca_vec = pucks.get_dataset(puck_name).get_gene_expression(gene='Hpca')

dge = pucks.get_dataset(puck_name).get_gene_expression()

puck.get_attribute(attribute='geneList')

aligned_locations = puck.get_aligned_locations(transformation='model_sift')

aligned_locations = puck.apply_save_transform(transformation='model_sift')