Figure 3

Inferred networks topologies under different levels of data noise. The different network topologies inferred under different levels of data noise. We show the best results obtained across the different sets of parameters from the LHS sampling method. a) The true segment polarity gene network; b) The network topology inferred under 0% of noise; c) The network topology inferred under 1% of noise; d) The network topology inferred under 5% of data noise. Solid red edges represent the inferred interactions that are not present in the real network, or that have the wrong direction (false positives), and dotted gray lines represent false negatives.