Figure 1

Parameters effecting thermodynamic consensus RNA structure predictions. As a basic parameter set we used equal base frequencies of 0.25, a transition/transversion rate ratio κ = 1, and the following tree ((A:0.09,B:0.09):0.09,(C:0.09,D:0.09):0.09) One parameter was varied at a time while others were kept constant. If necessary branch lengths were adjusted to keep a mean pairwise sequence identity (MPI) of 0.75 ± 0.01. 1000 alignments of length 80 were simulated under each condition. Cumulative histograms for the RNAalifold consensus folding energies are shown. Please note that we plot negative minimum free energies, i.e. higher values correspond to more stable folds. (A) Base frequencies were varied to get high and low G+C content. (B) Two specific dinucleotide frequencies were elevated 3-fold while the mononucleotide content was kept constant. (C) Branch lengths were equally scaled to produce alignments with lower or higher MPI identity than for the basic tree. (D) The transition/transversion rate ratio was varied. κ = 1 means equal rates, while κ > 1 gives more transition than transversions. (E) The alignment of size 80 was divided into a central block of 40 and two anking regions of 20. We set 100% conservation in the central block and low conservation in the anks (rate "high-low-high") and the other way round ("low-high-low"). The total average MPI was always 0.75. (F) We tested all possible topologies of this 4 taxa tree and adjusted the branch lengths to give a MPI of 0.75. For one given topology, all the branch lengths were of the same length.