Modelling Nonstationary Gene Regulatory Processes

Table 1

Predictive Probabilities: Macrophages data. Logarithmic predictive probabilities for the macrophage data: log_𝑒(𝑃(𝐃_‡|𝐃)) for BGe, BGM (as reported earlier [4]) and the new BGM_𝐷 model. The standard deviations of the logarithmic probabilities are given in brackets. We note that the BGM values could be confirmed by our reanalysis of the data: the deviations were smaller than one standard deviation.

		𝐃‡=𝐃TEST
𝐃=𝐃TRAIN	Model	CMV	IFN𝛾	CMV and IFN𝛾
CMV	BGe	—	−76.01 (±0.07)	−45.26 (±0.03)
	BGM	—	−63.63 (±0.02)	−33.80 (±0.38)
	BGM𝐷	—	−59.13 (±0.02)	−31.76 (±0.25)
IFN𝛾	BGe	−56.78 (± 0.05)	—	−57.30 (±0.05)
	BGM	−39.62 (± 0.02)	—	−42.69 (±0.11)
	BGM𝐷	−34.08 (± 0.14)	—	−39.11 (±0.10)
CMV+IFN𝛾	BGe	−37.76 (±0.08)	−69.19 (± 0.06)	—
	BGM	−21.67 (±0.33)	−53.26 (± 0.51)	—
	BGM𝐷	−18.58 (±0.09)	−51.11 (± 0.21)	—