Skip to main content
. 2016 Dec 15;17:532. doi: 10.1186/s12859-016-1391-0

Fig. 2.

Fig. 2

Workflow of the algorithm. The algorithm presents the iterative sampling of the model parameters in the Gibbs procedure. Parameter values (switch locations, heights, standard deviation) and fitted data are stored for each current model. The generated switch locations are accepted only if they produce a valid model. The parameters of the fitted model (switch heights and covariance matrix) are used to generate new sample values for the switch heights. Then, upon model fit, the fitted standard deviation is used to produce a sample value for the standard deviation. The next iteration with the updated model proceeds, new switch locations will be generated