Skip to main content
Figure 5 | BMC Bioinformatics

Figure 5

From: Partial inhibition and bilevel optimization in flux balance analysis

Figure 5

Effect of the precision parameter P (results from the reactions screening). (a) The plots show the number of solutions that are found for different values of τ (see legend) as function of P (color codes as in the following plots). As one can see, for τ > 0, increasing the precision P, it is nearly always more frequent to find a drug combination which induces the sought modulation. (b) The plot is a detail of the curve in panel (a) for the case τ = 0.5, and shows the frequency of the drug cardinality of the solutions (i.e. the number of drugs used in the solution). For larger P, the distribution is slightly shifted to higher cardinality (data for each τ are reported in Table 3). (c) Beside the number of solutions, higher precision produces more selective outcomes, i.e. with a lower side effect (as clearly shown by the mean values reported in the legend). The counting is performed over all different values of the threshold τ. (d) This plot shows the histogram of the values of the nonlinearity index η(h) (as reported in (11)) calculated for all the solutions of the screening (still regardless the value of τ). From the curves and from the mean values reported in the legend, it is possible to see that higher amount of nonlinearity are obtained when P is increased.

Back to article page