Figure 3.

Computational complexity of the GraphAlignment and G ræmlin algorithms. The scaling parameters estimated from the best power law fit of the data are given in the panels for the scenarios (i-iii). While the computational cost of GraphAlignment remains constant in all the scenarios, Græmlin’s performance deteriorates with addition of spurious weak vertex similarities in scenario (iii).

Kolář et al. BMC Systems Biology 2012 6:144   doi:10.1186/1752-0509-6-144
Download authors' original image