This is a 'best faith' effort at implementing SigProfiler as described in [1]. To my best knowledge, no one has attempted to reproduce signatures found in [1] using this repository. Note that the 'spherical k-means clustering' algorithm described in [1] is not the clustering algorithm used to find centroids in this implementation.
This repository was produced for the purpose of a class project and is released publically via the CRAPL license.
Please refer to the following examples for usagage:
- An example of training SigProfiler - train_sigprofiler.py
- Am example of accessing learned signatures -plot_err_vs_silhouette.py
pip install .
- Alexandrov, L. B., Nik-Zainal, S., Wedge, D. C., Campbell, P. J. & Stratton, M. R. (2013) "Deciphering signatures of mutational processes operative in human cancer." Cell Rep. 3, pages 246–259. doi:10.1016/j.celrep.2012.12.008