Abstract
Recovering deep phylogeny is challenging with animal mitochondrial genes because of their rapid evolution. Codon degeneration decreases the phylogenetic noise and bias by aiming to achieve two objectives: (1) alleviate the bias associated with nucleotide composition, which may lead to homoplasy and long-branch attraction, and (2) reduce differences in the phylogenetic results between nucleotide-based and amino acid (AA)-based analyses. The discrepancy between nucleotide-based analysis and AA-based analysis is partially caused by some synonymous codons that differ more from each other at the nucleotide level than from some nonsynonymous codons, e.g., Leu codon TTR in the standard genetic code is more similar to Phe codon TTY than to synonymous CTN codons. Thus, nucleotide similarity conflicts with AA similarity. There are many such examples involving other codon families in various mitochondrial genetic codes. Proper codon degeneration will make synonymous codons more similar to each other at the nucleotide level than they are to nonsynonymous codons. Here, I illustrate a “principled” codon degeneration method that achieves these objectives. The method was applied to resolving the mammalian basal lineage and phylogenetic position of rheas among ratites. The codon degeneration method was implemented in the user-friendly and freely available DAMBE software for all known genetic codes (genetic codes 1 to 33).
Highlights
In a multiple sequence alignment, there are historical signals, such as the number of nucleotide substitutions, that are typically proportional to the divergence time, and non-historical signals that are typically not proportional to the divergence time [1]
Nucleotide composition bias refers to the phenomenon in which distantly related taxa share similar nucleotide frequencies, leading to a spurious similarity between such taxa [8,9]
One could perform three different types of codon degeneration to alleviate the problems caused by composition bias and conflict signals between nucleotide and amino acid (AA) sequences
Summary
In a multiple sequence alignment, there are historical signals, such as the number of nucleotide substitutions, that are typically proportional to the divergence time, and non-historical signals that are typically not proportional to the divergence time [1]. Non-historical signals include compositional bias [2,3,4] and conflicting signals between codons and amino acids (AAs) [2,3,5,6]. Codon degeneration, when properly implemented, can minimize or eliminate these non-historical signals in aligned sequences [7]. I will outline these two sources of undesirable signals, detail a codon degeneration method, and apply the method to mammalian and avian mitochondrial sequences to resolve (1) the phylogeny of basal eutherian lineages and (2) the phylogenetic position of rheas among ratites. Nucleotide composition bias refers to the phenomenon in which distantly related taxa share similar nucleotide frequencies, leading to a spurious similarity between such taxa [8,9]. The problem is serious when one aims to construct a universal tree [10]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
