Home
REVOLVER is a tool for Transfer Learning (TL) in Cancer Evolution.
TL is a relatively new declination of Machine Learning in which we share information across learning tasks and domains to increase our performance. TL is an umbrella term for several paradigms; in REVOLVER what we do is actually Multi-task Learning.
REVOLVER correlates the task of selecting n trees from n patients for which sequencing data from one or more samples is available. The method, its mathematical ground and the pseudocode of the algorithm are described in the main paper.
- G. Caravagna, Y. Giarratano, D. Ramazzoti, I. Tomlinson, T.A. Graham, G. Sanguinetti, A. Sottoriva. Detecting repeated cancer evolution in human tumours from multi-region sequencing data. Nature Methods 15, 707–714 (2018).
If you are interested in general introduction to TL, you might start from a classic:
- Pan SJ & Yang Q (2010). A survey on transfer learning. IEEE Trans Know Data Eng, 22(10), 1345–1359.
Current version: released ~June 2018, codename "Haggis and tatties"
Author: Giulio Caravagna [@gcaravagna], Institute of Cancer Research, UK.
For any query about the tool, contact giulio.caravagna@icr.ac.uk; for requests of support open an issue here on Github
# Required external packages available on CRAN:
# "cluster", "crayon", "dendextend", "dynamicTreeCut",
# "doParallel", "foreach", "igraph", "parallel"
# Required external packages available on GitHub: "caravagn/pio"
devtools::install_github("caravagn/revolver")