Computational study of noise in a large signal transduction network.
Bottom Line: Although noise is often perceived as a disturbing factor, the system might actually benefit from it.In order to understand the role of noise better, its quality must be studied in a quantitative manner.We concluded that basic frequency domain techniques can be applied to the analysis of simulation results produced by the Gillespie stochastic simulation algorithm.
Affiliation: Department of Mathematics, Tampere University of Technology, P.O. Box 553, 33101 Tampere, Finland. email@example.comShow MeSH
Related in: MedlinePlus
Mentions: The dependence between the system volume and the quality of noise can also be studied in the frequency domain. In general, the frequency domain analysis of the simulation results shows that most of the power in these noise processes is on the low frequencies. The quality of the PSD estimates depends somewhat on the stationarity of the noise processes studied. In order to illustrate the volume dependence of the quality of noise in the frequency domain, we have selected two model species: phospholipase C (PLC), and calcium phospholipase C complex (CaPLCcomplex). Illustrative realizations of these species are shown in (Figure 5A and 5B). The realizations are simulated in four different system volumes, 5 × 10-16 l (blue), 3.5 × 10-15 l (red), 10-14 l (green), and 10-13 l (black). The realizations show how the discrete nature of reactions plays an important role in the smallest volume and how the strength of noise gets smaller when the system volume increases. The effects of noise are however still detectable also in the largest volume. By taking a look at the PLC and CaPLCcomplex noise processes in (Figure 5A and 5B), it is easy to conclude that they are somewhat different. Therefore it is interesting to see how their behavior differs in the frequency domain. We estimated PSDs of the noise in the PLC and CaPLCcomplex concentrations using 10 second realizations of these processes. The frequency content of these noise processes as a function of volume can be seen in (Figure 5C and 5D). The noise in both species seems to have most of its power on the lower frequencies. Similar behavior was also observed in other species in the network. The PSDs for PLC and CaPLCcomplex seem to have different shapes (see Figure 5C and 5D). The noise in PLC realizations clearly has the dominating power on very low frequencies whereas the frequency content of noise in CaPLCcomplex realizations is distributed more uniformly on low frequencies. An interesting observation also is that the shapes of PSDs for species PLC and CaPLCcomplex seem to be same in all studied volumes. This can be seen clearly in Figure 6, where four PSDs in different volumes for PLC and CaPLCcomplex are plotted using the log-log-scale.
Affiliation: Department of Mathematics, Tampere University of Technology, P.O. Box 553, 33101 Tampere, Finland. firstname.lastname@example.org