A Bayesian framework for cancer detection

Blood-based tests are among the most-promising technologies for minimally invasive detection of cancer at early stages. These tests are based on the analysis of cell-free DNA, fragments of DNA released by cells that can be found in the blood and other body fluids. When developing a cancer mass, some of these fragments will be released by cancerous cells. DNA from these cells will differ by other fragments because of acquired mutations.

The main challenge in developing cancer detection tools at early cancer stages, is in that the proportion of DNA fragments from cancer cells is extremely low (even less than 1%). Advanced statistical methods are necessary to distinguish the signal from the noise.

This is mostly work in progress, but the statistics behind this project encompass Hidden Markov Models, Bayesian testing procedures and the use of biologically informed priors.