Installation

Setup snakemake workflow

This guide will describe how to set up an environment using micromamba, but conda or mamba are also fine. Micromamba seems to be the fastest. We’ll add some specific advice for setting it up on a VSC account.

To install micromamba (Reference):

# If on the VSC:
cd $VSC_DATA

# Download installation script
curl -L micro.mamba.pm/install.sh > install_micromamba.sh

# Ideally, skim the script to check it for issues, then:
bash install_micromamba.sh

This should show you a step-by-step wizard:

Micromamba binary folder? [~/.local/bin]
Init shell (bash)? [Y/n]
Configure conda-forge? [Y/n]
Prefix location? [~/micromamba] -> You might want to change this, see below

Important for the VSC: When asked for a “root prefix”, you can choose:

  • .micromamba, without the ~ at the beginning. This will create one folder named .micromamba where you currently are, which contains the default environment. When you create a new environment while inside a project, it will create a new .micromamba folder inside that project for your new environment.

  • /some/absolute/path/micromamba: This will create one directory that will hold all your environments. On the VSC, you can echo $VSC_DATA/micromamba and then copy that.

By default, micromamba creates environments in the home directory. You don’t want that on the VSC or you’ll run out of quota. The .micromamba directory is gitignored in the protein-runway project, so it’s a fine choice. It’s also okay to choose $VSC_DATA/micromamba or whatever you like inside of $VSC_DATA, just make sure to expand it beforehand.

Set up and activate a protein-runway environment from the env file in the repo:

micromamba create -f micromamba_env.yml
micromamba activate protein-runway

Arguments:

  • -n protein-runway: Name of the environment

  • -c conda-forge: Source for installing dependencies

  • python=3.11: The specific version that is required for the project

Install dependencies:

pip install -r requirements.txt

This will likely still show some dependency errors, but they don’t seem to stop the pipeline from working. An alternative setup for the repository might be to check out the individual projects’ requirements in separate virtual environments or even micromamba environments for better isolation.

Run snakemake workflow

Before running any code, we need to activate the micromamba environment:

micromamba activate protein-runway

The input data consists of two files per protein from the Zenodo project we have been testing with (https://zenodo.org/records/4650406):

  • Trajectory file from Trajectory_snapshots_<code>_<code>.zip:

    • e.g. 1fuu_noPTM_10-20ns_100snap.trr

    • Location: 01_input/traj/

  • Topology file from Topology_stripped.zip:

    • Example: 1fuu_noPTM_complex.top

    • Location: 01_input/top/

At that point, snakemake should build all necessary output files for these proteins into 03_output.

Build blender extension

The easiest way to get the blender extension is to use the version that is built by the github actions pipeline, an archive attached to the “Releases” of the project: https://github.com/AndrewRadev/protein-runway/releases.

If you’d like to set it up yourself from the source, you need to install packages and zip them. From the root of the project:

pip wheel MDAnalysis -w ./blender/extension/wheels/

Then zipping the extension can be done by running:

bash blender/build.sh

But this assumes you’re on Linux or Mac, so if you’re on Windows, you can just take the contents of the “blender/extension” folder and zip them manually.

Either way, you can then install the addon from Edit > Preferences > Get Extensions. Then, in the right-hand corner menu, Install from Disk.

What blender does is it just unzips that into a local folder. On Linux, that folder is ~/.config/blender/4.2/extensions/user_default. So if you are actively changing the extension, you can symlink the local folder directly there, so that when you edit the code, you can just re-launch blender.