A Bayesian approach for parameter estimation in the extended clock gene circuit of Arabidopsis thaliana

BMC Bioinformatics

Table 1 Summary of convergence diagnostics in parameter and data space

Experiment	Euclidean distance from true values at start ED₀	Species Analysis PSRF computed from sum-of-squares trace plots in data space			Parameter Analysis PSRF computed from MCMC chains in parameter space			Euclidean distance between true values and posterior mean ED_m	%True parameter values lying outside 5^th - 95^th percentile of posterior distribution
		%Species PSRF <1.1	Mean PSRF	MPSRF	%Parameters PSRF <1.1	Mean PSRF	MPSRF
1. No perturbation	0	100	1.01	1.81	99	1.02	2.69	0.25	0
2. Perturbed variance = 0.005²	0.06	100	1.01	1.77	95	1.03	3.38	0.30	0.8
3. Perturbed variance = 0.01²	0.12	97	1.01	1.80	90	1.03	3.07	0.22	1.5
4. Perturbed variance= 0.015²	0.20	97	1.01	1.85	92	1.03	3.68	0.30	1.5
5. Perturbed variance= 0.02²	0.24	97	1.02	2.12	96	1.03	3.47	0.33	0.8
6. Perturbed variance= 0.025²	0.30	93	1.03	2.11	68	1.09	3.37	0.32	3.0
7. Perturbed variance = 0.03²	0.32	100	1.01	1.74	100	1.01	2.85	0.43	3.8
8. Perturbed variance = 0.035²	0.38	87	1.05	2.50	62	1.11	3.82	0.37	6.8
9. Perturbed variance = 0.05²	0.51	57	1.10	2.26	45	1.21	3.48	0.49	10.5
10. Perturbed variance = 0.1²	0.98	33	1.35	4.47	24	1.41	5.79	0.59	11.3
11. Gamma(1,2)	87.66	7	1.98	8.47	18	2.14	10.29	0.70	21.1
12. Gamma(2,4)	39.86	7	2.31	9.14	5	2.04	10.73	16.03	21.1
13. Gamma(2,4)	88.12	3	2.36	5.81	2	1.82	6.89	40.90	15.8

ISSN: 1471-2105