Installation

Docker image

A GPU-enabled docker image is available on the Google Artifact Registry here:

us-central1-docker.pkg.dev/broad-dsde-methods/cellarium-ai/cellmincer:0.1.0

Installing from source

We recommend installing CellMincer within a dedicated Python environment created via conda or venv to prevent potential dependency conflicts.

Create a conda environment and activate it:

$ conda create -n cellmincer python=3.12
$ conda activate cellmincer

Install with pip from Git repository:

(cellmincer) $ pip install git+https://github.com/broadinstitute/CellMincer.git

Because CellMincer uses GPU computing for both training on and denoising voltage imaging, you may wish to confirm that your machine has GPUs with appropriate drivers installed and is CUDA-enabled (e.g. check that torch.cuda.is_available() returns True).