Table 1. Comparison of crosstalk levels between the different variants of the model.

Model	Crosstalk (at baseline parameters)	Remarks
Basic model (activators-only)	0.23
Basic model (repressors-only)	0.23
Mixed model (activators+repressors)	0.14	2,000 genes expressed in 20% of the env., 3,000 genes in 70%
Genes of unequal importance	0.31	10% of the genes are important and penalized 10 × the ‘normal' rate
		The resulting error per important gene decreases to 0.1, but for the other genes increases to 0.33.
Unequal weights for the two error types	0.17	b=0.5, weight of erroneously active genes is half that of genes that are erroneously inactive
Each TF regulates exactly Θ=10 genes	0.08	Also holds for P(Θ)∼Poisson
Activators+global nonspecific repressor	0.23	Cannot reduce crosstalk
Activators+specific repressors (non-overlapping)	0.2
Activators+specific repressors (overlapping)	0.15
Perfect AND-gate combinatorial regulation	0.07	Uses only TF species
Generic cooperativity	0.064	For example, dimerization, direct TF–TF contacts, TF/nucleosome competition and so on. Two bindings sites, each of length L=10
Cooperativity exclusive to cognate binding	0.006	Currently unknown molecular mechanisms, two bindings sites, each of length L=10.

Baseline parameters are: Q=2,500, M=5,000, log(S)=−10.5—equivalent to a model where binding sites for distinct TFs are different from each other in at least 2 bp (d_min=2) with binding sites of L=10 bp and binding energy per mismatch.