A unified framework for estimating parameters of kinetic biological models

BMC Bioinformatics

Table 1 Parameter estimation results using the CSUKF, parameter ranking and profile likelihood analysis from the sugarcane model

		CSUKF
Parameter name	Actual value	Mean	Std. Dev.	Orthogonal ranking	Proflile likelhood analysis
V_max6r	0.2	0.34	0.670	1	Practically NI
K_m6Suc6P	0.1	5.97	4.580	2	Structurally NI
K_i6UDPGlc	1.4	0.32	0.400	3	Identifiable
K_i1Fru	1	1.00	0.010	4	Identifiable
K_i3G6P	0.1	0.67	1.460	5	Practically NI
K_m6UDP	0.3	4.73	3.450	6	Structurally NI
V_max11	1	0.28	0.190	7	Identifiable
K_i6Suc6P	0.07	0.45	0.770	8	Practically NI
K_i2Glc	1	1.00	0.009	9	Identifiable
K_i4F6P	10	0.63	0.850	N.I.	Practically NI
K_i6F6P	0.4	0.65	1.060	N.I.	Practically NI
K_m11Suc	100	21.43	21.820	N.I.	Practically NI

N.I. - Not Identifiable.
The mean and standard deviation of the estimated parameters are calculated from 50 repetitions. The ranking is chosen based on the weighted average ranking from each of the 50 runs. The profile likelihood analysis determines all non-identifiable parameters and classifies the non-identifiability as practical or structural. In each repetition the parameters are randomly initialized to values between 0 and 1.

ISSN: 1471-2105