====== MuSpAn ======

From https://www.muspan.co.uk:

> MuSpAn is a multiscale spatial analysis toolbox for analysing complex imaging data. It uses cutting-edge mathematical and statistical approaches to data analysis to provide the most comprehensive spatial analysis available.
>
> MuSpAn is being continually expanded, with a team of quantitative researchers constantly developing new methodology to tackle problems in analysing spatial transcriptomics, multiplex immunohistochemistry, imaging mass cytometry and more.

   * See more information: https://www.muspan.co.uk/
   * MuSpAn documentation: https://docs.muspan.co.uk/latest/Documentation.html

----

===== Installing MuSpAn on Comet =====

First you will need to complete the [[ https://www.muspan.co.uk/get-the-code|MuSpAn application form]] and confirm that the software is only to be used for academic purposes. The username and password you receive will be used later to install the software, regardless of which method you choose.

After getting the MuSpAn download credentials, you have //three// possible options to install the software on Comet.

===== Option 1. Using Python Module + Pip =====

Use the basic Python module as included on Comet and the provided ''pip'' command to do a //user// level install of MuSpAn into your own local Python modules folder (this resides within your ''$HOME'' directory under ''$HOME/.local/lib/python*''):

<code lang=bash>
$ module load Python/3.12.3
$ pip install --user https://docs.muspan.co.uk/code/latest.zip
<prompted for MuSpAn username>
<prompted for MuSpAn password>
$
</code>

To test:

<code lang=bash>
$ module load Python/3.12.3
$ python -c "import muspan as ms; print('MuSpAn module available')"

MuSpAn module available

$
</code>

To run in your own scripts afterwards:

<code lang=bash>
$ module load Python/3.12.3
$ python <script_name.py>
</code>

Although easiest, the disadvantage is that MuSpAn (and all of its dependencies installed by ''pip'') are always present, even if you decide you need a different //pandas// or //numpy//. We therefore recommend **Option 2**, or **Option 3**, below.
----
===== Option 2. Using Python Module + Virtual Env =====

Create a new Python virtual environment using the built-in Python module commands. The folder ''./muspan_venv'' will be created in whichever directory you run the commands below:

<code lang=bash>
$ module load Python/3.12.3
$ python -m venv muspan_venv
</code>

Activate that new environment:

<code lang=bash>
$ cd muspan_venv
$ source bin/activate
</code>

Now use ''pip'' to install MuSpAn in that environment:

<code lang=bash>
$ pip install https://docs.muspan.co.uk/code/latest.zip
<prompted for MuSpAn username>
<prompted for MuSpAn password>
$
</code>

To test:

<code lang=bash>
$ module load Python/3.12.3
$ cd muspan_venv
$ source bin/activate
$ python -c "import muspan as ms; print('MuSpAn module available')"

MuSpAn module available

$
</code>

To run in your own scripts afterwards:

<code lang=bash>
$ module load Python/3.12.3
$ cd muspan_venv
$ source bin/activate
$ python <script_name.py>
</code>

This keeps MuSpAn and all of its extra dependencies contained within the ''muspan_venv'' folder so that it does not conflict with any other Python software you have installed or use.

----
===== Option 3. Using Conda/Mamba Environment =====

Use [[advanced:software:miniforge|Miniforge]] to create a new **conda** (or **mamba**) environment. This will create the new folder ''./muspan_env'' within ''$HOME/miniforge3/venvs/'':

<code lang=bash>
$ module load Miniforge
$ conda create -n muspan_env -c conda-forge
</code>

Activate the new environment:

<code lang=bash>
$ conda activate muspan_env
</code>

Ensure the new environment has ''pip'' installed:

<code lang=bash>
$ conda install pip
</code>

Then use pip from the environment to install MuSpAn.

<code lang=bash>
$ pip install https://docs.muspan.co.uk/code/latest.zip
</code>

Test MuSpAn is installed:

<code lang=bash>
$ python -c "import muspan as ms; print('MuSpAn module available')"

MuSpAn module available

$
</code>

You can then use MuSpAn afterwards in your own Python scripts by running:

<code lang=bash>
$ module load Miniforge
$ conda activate muspan_env
$ python <name_of_script.py>
</code>

----

[[:advanced:software|Back to Software]]