Two existing papers [2], [11] propose univariate feature selection techniques to build maximum likelihood estimate based nearest-centroid classifiers, the latter of which is called the "Classification to Nearest Centroids" (Clanc) method.
Aligned clusters were concatenated to generate super-alignments that were used to build maximum likelihood phylogeny tree.
MEGA v5.1 was used to find the most appropriate substitution models and to build maximum likelihood (ML) trees for the transposons and exons.
PHYML [ 36] was used to build maximum likelihood trees, assuming a JTT model of substitution and non-parametric bootstrapping (100 replicates).
The multiple alignment of the tubulin/FtsZ superfamily (see Additional File 2) was employed to build maximum likelihood phylogenetic trees using FtsZ proteins as the outgroup.
The best-suited nucleotide model for each alignment was determined among 88 possible models following the Akaike Information Criterion [ 58] and the Bayesian Information Criterion [ 59] (whenever the two criteria disagreed, the more parameter-rich model was selected) as implemented in jModelTest [ 60], and used to build Maximum Likelihood (ML) phylogenetic trees with PhyML 3.0 [ 61].
To reconstruct the evolutionary scenario for viral ligases, we used multiple alignments of the NAD-dependent and ATP-dependent ligases that included the respective protein sequences from the NCLDV, other viruses, and representative archaea, bacteria, and eukaryotes (see Additional File 1 and Additional File 2, respectively), to build maximum likelihood (ML) phylogenetic trees.
Next, we built maximum likelihood trees using fasttree (Price et al., 2009), collapsed zero-length branches into polytomies, and ranked external and internal nodes using the LBI.
Using posterior means derived from these Bayesian models we build a broader likelihood model to predict growth and survival of all species under a range of environmental and ontogenetic conditions, to identify relationships between traits and demography, and quantify the role of environmental and ontogenetic context.
Various subgroups were then selected and used to build further maximum likelihood trees as described above.
