- python 3.8.3 , pandas 1.2.4 , numpy 1.21.2, matplotlib 3.4.2, sklearn 0.24.2
- archetypes.py : open source software for archetypal analysis downloadable at https://doi.org/10.25919/5d3958889f7ff
- SigProfilerExtractor : open source software for mutational signature extraction downloadable at https://github.com/AlexandrovLab/SigProfilerExtractor
This repository contains the following folders:
- utils: COSMIC signatures profiles and usefull python functions
- notebook: 2 Jupyter notebook to reproduce paper analysis
- synthetic_catalogues: synthetic catalogues generated with SigsPack for each scenario
- archetypal_profiles: the archetypal profiles identified
-
Archetypal Analysis.ipynb can be executed after the installation of archetypal analysis package archetypes.py
-
De-novo Extraction.ipynb calculate metrics for each scenario after de novo extraction with SigProfilerExtractor. Given the number of simulation and the size of the results we could not upload all the results; Hence to run the notebook is necessary to first perform de novo extraction for each scenario with SigProfilerExtractor executing run_SigProfilerExtractor.py. The computation is really expensive and it may takes some times. Results may vary due to the stocasthic nature of the noise that SigProfilerExtractor uses before perfoming the repeated NMF.