This repo provides an implementation of the Probabilistic Image-based Spatial Transcriptomics Decoder (PoSTcode), which is a method for decoding image-based spatial transcriptomics based on a re-parametrised matrix-variate Gaussian mixture model, implemented using stochastic variational inference in pyro.
To get started, please explore the Jupyter Notebook example_iss_tile.ipynb, which provides an example of how the code can be used to decode an ISS tile from the mouse brain dataset stored locally and takes less than 1 minute to run.
The code has been tested with python 3.6.12 and its requirements can be fulfilled (in less than 5 minutes) by running
python3 -m pip install -r requirements.txt
python3 -m pip install -e .
The code is written by Milana Gataric (milanagataric at gmail.com). If you use this code, please cite the original paper where it was introduced: PoSTcode: Probabilistic image-based spatial transcriptomics decoder.