Computational protocol: The molecular signal for the adaptation to cold temperature during early life on Earth

Similar protocols

Protocol publication

[…] Boussau et al. [] built a concatenate of small- and large-subunit rRNAs from 456 organisms (2239 sites) and used the sites restricted to stem regions (1043 sites) to infer the ancestral G + C contents over the tree of life. From these alignments, we selected 125 species covering a broad taxonomic diversity without redundancy in the taxonomic sampling. Regarding the concatenation of proteins, the 56 gene families and 30 species considered in Boussau et al. [] were used here, and increased to 38 species, with the addition of Archaea species in particular, which were poorly represented in the first set of species. We reconstructed ML phylogenetic trees for rRNAs (on the 2239 sites dataset) and proteins with PhyML []. A three-domain tree was obtained and the root was placed on the branch between the ancestors of Bacteria and Archaea/Eukaryotes. As in [] and [], the branch-wise equilibrium frequencies were estimated along these universal phylogenetic trees. The stem dataset was analysed with the BppML program [] assuming a discrete gamma distribution with eight categories to model rate variation among sites and the non-homogeneous Galtier & Gouy (GG) substitution model []. The GG model specifies branch-wise equilibrium G + C contents, as well as an independent G + C content at the root. For proteins, we used a new branch-wise non-homogeneous model implemented in the maximum-likelihood (ML) framework, named COaLA [] that we recently designed. See the electronic supplementary material for a description of the COaLA model and an evaluation of the fit to data of the non-homogeneous models in comparison with homogeneous models. [...] OGT highly correlates with the G + C content of the stem regions of rRNAs (ρ = 0.76, p-value < 0.001; see the electronic supplementary material, figure S2) and with the second axis of the COA computed on amino acid compositions of the protein dataset restricted to prokaryotic species (ρ = 0.88, p-value < 0.001; see the electronic supplementary material, figure S3). We controlled for phylogenetic inertia with the phylogenetic independent contrast approach [] using the R package APE [] and observed that those correlations were still strongly significant. Linear regressions between OGTs and compositions were then computed to obtain the molecular thermometers. […]

Pipeline specifications

Software tools PhyML, Coala, APE
Applications Phylogenetics, Population genetic analysis