Fig 22. Simulation of a circiadian clock.
(A) Result of a stochastic simulation of a circadian clock model with NFsim. Copy numbers of molecules A and R show periodic oscillation. A low pass frequency filter was used to smooth the values of A and R. The reason for the smoothing was to get a numerical value related to the actual peak. The peaks from low-pass filtered data do not represent true average peaks but can be used as a proxy to obtain the time of a peak for comparison with other simulation methods. (B) The error bars capture the mean and standard deviation of the low pass filtered peak times for different variants of the model and simulation algorithms. Each of the variants was run 1024 times. It is evident that the SSA, the NFsim, and the MCell4 model variants with a fast diffusion constant, D = 10−5 cm2/s, give essentially the same results. The hybrid MCell4 model with the slower diffusion constant, D = 10−7 cm2/s, shows faster oscillation than the non-spatial models run with SSA and NFsim, and the MCell4 variants with faster diffusion. The pure particle-based MCell4 model with D = 10−7cm2/s shows the fastest oscillations.