Skip to content

Discovering biomarkers for meningitis by multivariate integration of multi-omics and clinical data

License

Notifications You must be signed in to change notification settings

krassowski/meningitis-integration

Repository files navigation

Meningitis omics integration with PLS, O2-PLS and LASSO

Build Status

Scope

Data: Transcriptomics (mRNA, gene-level) and proteomics (SOMAScan) integration + clinical data including survival, meningitis subtype and HIV status.

For the list of analyses, please see the Research_plan.md

Code organisation overview

Code is organised into:

  • Jupyter Notebooks (.ipynb) containing high-level analysis code, table summaries and visualisations. The notebooks can be viewed in the browser on the GitHub page of this repository.
  • Implementation details are stored in:
    • Python modules (.py)
    • R scripts (.R)

Most of the notebooks use the Python kernel, but some also contain R code. The R code in Python notebooks is marked by the %%R at the beginning of the cell (thanks to rpy2). An exclamation mark indicates a bash command, e.g. !ls.

The generic utilities and helper functions are stored in the helpers directory.

Automation for reproducibility

The order of notebooks execution is described by the workflow rules in the pipeline.py file. Following the installation with:

# download the code
git clone git@github.com:krassowski/meningitis-integration.git
# enter the directory (if you use conda/virtualenv, activate it now)
cd meningitis-integration
# install Python and R dependencies
./setup.sh
# download the data
./download.sh

execute nbpipeline to reproduce all results of our study with:

# -i will generate an interactive graph featuring reproducibility reports
PYTHONPATH=$(pwd):$PYTHONPATH nbpipeline -i

The analyses plot

For interactive plot use -i (as shown above), for static use -g:

nbpipline pipeline.py -g

Append -n switch to skip execution of the pipeline and generate minimal plots (without reproducibility reports and code analysis):

nbpipline pipeline.py -i -n

Version compatibility

Developed and tested with Python 3.7.6, R 3.6.3 and Ubuntu 19.10.

All dependencies are declared in:

  • requirements.txt for Python,
  • install.R for R

About

The code in this repository was written as a part of MRes research project at Imperial College London.

About

Discovering biomarkers for meningitis by multivariate integration of multi-omics and clinical data

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages