The Shimodaira-Hasegawa test (implemented in PAUP* with 1000 replicates) was then used to compare the resultant log likelihood scores of each constrained tree to the original tree.
The differences in likelihood value between the optimal tree and each constrained tree were calculated and the significance evaluated using the Shimodaira-Hasegawa test [54] thus quantifying the difference in phylogenetic signal between different loci along the expression site.
Thus mRNA and protein synthesis are each constrained to have maximum values of 1.0 s-1.
In Figure 8, we plot the absorption wavelength having the highest oscillator strength (among the 10 lowest excitations) for each constrained geometry.
The per-site log likelihood scores were calculated for the best-scoring ML trees under each constrained topology using the -f g option in RAxML.
To constrain Renocera + Anticheta as a lineage outside of the other tribes, the Sciomyzini, Tetanocerini and Renocera + Anticheta were each constrained to be monophyletic with three separate constraint statements.
The means of the growth factors were allowed to vary across the latent classes, but the variances and covariances of the growth factors were each constrained to be equal across the latent classes.
Each constrained tree was based on an initial guide tree with a single internal branch, generated in MacClade version 4.08 [ 56] or in Mesquite version 2.75 [ 57] (Additional file 6).
The Asian and Australo-Papuan clades were each constrained to be monophyletic and the age of the root, the age of the most recent common ancestor (tMRCA) for Rattus, was assigned a lognormal prior based on the empirical results from the WG analysis.
Due to the pruning of the search space, each constrained docking run is around 20 to 30 times faster than the unconstrained search, which allows us to screen the highest scoring 100 contacts predicted by the classifier in approximately ten to fiteen hours per complex, on average (279 core hours total for the bound dockings, 370 hours for the unbound dockings).
We generated 10,000 bootstrap replicates of the site likelihoods for each constrained tree using CONSEL [ 63], and ranked alternative hypotheses using the weighted Shimodaira-Hasegawa test (WSH [ 63]), the approximately unbiased test (AU [ 64]), and the Bayesian Information Critierion approximation for posterior probability (BIC [ 63]).
