EMu: Expectation-Maximisation inference of mutational signatures

EMu is software for inferring the mutational signatures present in a number of cancer mutation sets.

Several independent mutational processes jointly produce the observed spectrum of mutations in a number of comparable tumours. EMu tries to find the number of processes present in a data set and their individual mutational signatures. These signatures contain information about the biological, physical or chemical processes active in cancer.

EMu can also be used to localise regions in a cancer genome where a given set of mutational processes is active. EMu exploits the fact that the outcome of a mutational process not only depends on its signature, but also on the sequence on which it acts.

[Genome Research Limited]


The software is provided as a C++ source code. It requires a C++ compiler and a recent version of the GNU Scientific Library (gsl-1.14 or later).


  • EMu: probabilistic inference of mutational processes and their localisation in the cancer genome

    A Fischer, CJR Illingworth, PJ Campbell and V Mustonen


The latest version of the software can be found at:

In the same directory, interactive visualisations of the localisation of mutational processes in 21 breast cancers can be found (requires Mathematica.s free cfd-player, http://www.wolfram.com/cdf-player/).


If you have any questions or feedback, please contact Andrej Fischer.

* quick link - http://q.sanger.ac.uk/emu