neonine2 / morpheus-spatial

Counterfactual generation of tumor perturbations from multiplexed tissue images
19 stars 2 forks source link
Morpheus logo

Morpheus is an integrated deep learning framework that takes large scale spatial omics profiles of patient tumors, and combines a formulation of T-cell infiltration prediction as a self-supervised machine learning problem with a counterfactual optimization strategy to generate minimal tumor perturbations predicted to boost T-cell infiltration.

Graphical summary of the Morpheus framework

Update: We are currently working on tutorial notebooks for running optimization on Slurm with fan-out across multiple CPU nodes, stay tuned!

Getting Started

Prerequisites

Note numpy 2.0 or above not currently supported

Installation

Option 1: Using pip (PyPI)

Run the following in the command line

pip install morpheus-spatial

Option 2: From Source

To install Morpheus from source, clone the repository and install the dependencies:

git clone https://github.com/neonine2/morpheus-spatial.git
cd morpheus-spatial
pip install -r requirements.txt
pip install .

Tutorial

See tutorial_notebook.ipynb for a complete, self-contained workflow on using Morpheus to generate therapeutic strategies.

Known Issues

OpenMP Conflicts on macOS

Some users may encounter warnings about conflicting OpenMP libraries. If you see a warning about Intel OpenMP and LLVM OpenMP being loaded at the same time, please see https://github.com/joblib/threadpoolctl/blob/master/multiple_openmp.md for more information and possible workarounds.

Repository Structure