PAMOGK

Installation

Getting the Code Base

You can either checkout the code base through git:

git clone https://github.com/tastanlab/pamogk.git

or you can download the master branch from:

https://github.com/tastanlab/pamogk/archive/master.zip

Python Environment

Project mainly uses python, but some of the plot generation codes use matlab or output html files (plotly.js). Also some libraries might be dependent on operating system. In our tests we used Debian 12.04LTS and macOS Catalina.

First install pipenv using the instructions given:

https://github.com/pypa/pipenv#installation

We are using pipenv because of version locking and predictive builds.

Setting up MOSEK

We are using MOSEK Optimizer API for optimizing view weights, and in order to use MOSEK you will need a license file. MOSEK provides free certificates for academic purposes and free trial certificates for Commercial usages. You can see check the link below for information on how to acquire an academic license:

https://www.mosek.com/products/academic-licenses/

License File Placement

You can either place the license file under user's home directory in a folder named mosek e.g:

$HOME/mosek/mosec.lic # *nix/macOS users
%USERPROFILE%\mosek\mosek.lic # windows users

Other option is to set the MOSEKLM_LICENSE_FILE environment variable to your license file location.

We also look for the license file paths in some custom paths (by overwriting MOSEKLM_LICENSE_FILE if not present). You can check pamogk.config.MOSEK_SUPPORTED_PATHS to see a list of supported paths.

For more information:

https://docs.mosek.com/9.2/install/installation.html#setting-up-the-license

Dependencies

Install dependencies with:

pipenv install

Required Environment Variables

You can add a .env file that will be loaded by pipenv

Also please don't forget to export MOSEKLM_LICENSE_FILE variable as well if you are using a path other than $HOME/mosek/mosek.lic:

# required for experiment running
PYTHONPATH=${PYTHONPATH}:.
MOSEKLM_LICENSE_FILE=$MY_MOSEK_LICENSE_FILE_PATH

Folder Structure

• pamogk: package root folder
  • kernels: Kernel methods to calculate distance between patients. Will try to extend these!
  • data_processor: Data processors/converters used to convert different forms of data to usable forms. Will try to extend these!
  • gene_mapper: Uniprot/Entrez gene mapping matcher for different datasets using different gene IDs. Will try to separate to its own package or along with other tools.
  • pathway_reader: https://.ndexbio.org CX format, and http://kegg.jp KGML format reader and converters. Will try to separate to its own package or along with other tools.
• data: Data files that are both raw or generated by the project. Keeping intermediate files saves a lot of computation time where applicable. (not included in package)
• experiments: Experiments done on real data. (not included in package)

Experiments

Running Experiments

You should run experiments either by starting a pipenv environment shell by:

pipenv shell

or through pipenv with:

pipenv run experiments/pamogk_all_exp_1.py

This ensures that experiments are run with correct environment.

Experiment Parameters

Experiments are code entry points that have arguments that change per experiment. You can see their arguments by running them with --help parameter. e.g:

./experiments/pamogk_all_exp_1.py --help

and you should see help as such:

usage: pamogk_all_exp_1.py [-h] [--rs-patient-data file-path]
                           [--rp-patient-data file-path]
                           [--som-patient-data file-path]

Run PAMOGK-mut algorithms on pathways

optional arguments:
  -h, --help            show this help message and exit
  --rs-patient-data file-path, -rs file-path
                        rnaseq pathway ID list
  --rp-patient-data file-path, -rp file-path
                        rppa pathway ID list
  --som-patient-data file-path, -s file-path
                        som mut pathway ID list