Setup

First clone the repository locally, including the ProteinCartography submodule.

git clone --recurse-submodules [email protected]:latchbio/wf-protein-cartography-single.git

Next, you can register the pipeline normally by running

latch register -y .

Pipeline Summary

The workflow defined here calls the ProteinCartography snakemake workflow via subprocess, on a single machine configured to use 2 vCPUs and 4 GiB RAM. Environment management is done via mamba.

The entrypoint code is in wf/entrypoint.py, with parameter metadata stored in wf/metadata.py.

The code is run inside of a docker image which is fully specified in Dockerfile. The snakemake binary is installed inside of the cartography_tidy env. This env is used to call the pipeline.

Currently the pipeline recreates every environment it needs each time it runs. A potential enhancement would be to pre-build all of these environments in the host docker image and reference them by name. I did not do this here as that would require modifying the ProteinCartography pipeline directly.