Automated Bayesian model development for frequency detection in biological time series

BMC Systems Biology

Table 1 Model ratios for number of frequencies in the data

Case	N _ω	Models	Model ratio
A	1	H_1ω/H_2ω	66936
B	2	H_1ω/H_2ω	0
B	2	H_2ω/H_3ω	686

Case A has only one frequency, N_ω = 1, and case B has two, N_ω = 2. The model ratio of a model including one frequency, H_1ω, and a model with two, H_2ω, is well above 1 for case A, showing that the simpler model should be chosen. For case B, the first ratio is close to 0, showing that the more complex model with two frequencies is far more likely. Values below 10^-8 are listed as 0. The second ratio is above 1, preferring the model with two frequencies to the one with three.

ISSN: 1752-0509