The effect of alignment uncertainty, substitution models and priors in building and dating the mammal tree of life.

摘要: The flood of genomic data to help build and date the tree life requires automation at several critical junctures, most importantly during sequence assembly alignment. It is widely appreciated that automated alignment protocols can yield inaccuracies, but relative impact various sources error on phylogenomic analysis not yet known. This study employs an updated mammal set 5162 coding loci sampled from 90 species evaluate effects uncertainty, substitution models, fossil priors gene tree, divergence time estimation. Additionally, a novel coalescent likelihood ratio test introduced for comparing competing trees against given trees. aligned DNA sequences were trimmed filtered using trimAL two filtering protocols. final dataset contains 4 sets alignments - before trimming, after by recently proposed pipeline, further ML each locus with concatenation tree. Our analyses suggest average discordance among significantly smaller than estimated or different models. There no significant difference times However, dates trimming are more recent those trimming. results highlight uncertainty choice models have little topologies yielded methods estimation, whereas they influential made concatenation. Given calibration scheme clock estimates robust removing deemed problematic algorithms lead dates. Although prior important in Bayesian this driven primarily data.