Phylogenetic Analysis May be Misleading Due to Compositional Bias

Phylogenetic Analysis May be Misleading Due to Compositional Bias

In this paper, Foster et al* looked at DNA composition bias and how it adversely affects phylogenetic analysis. That is, phylogenetic analysis produces false results by grouping unrelated taxa. This has been shown to be the result of compositional bias i.e. AT/GC-rich components of the genome. With these biases, it is clear how errors would occur in phylogenetic analysis—they are being grouped together based on biased areas of the genome. For example, the nematode and honeybee were grouped together even though they are obviously biologically unrelated. They also mention how bias in the DNA can affect protein bias.

In their experiment, Foster et al performed phylogenetic analyses of the protein coding sequences from the mitochondria. The specie’s genes selected for analysis were known to have varying amounts of both DNA and amino acid composition bias and therefore, they are ideal for observation of the effect(s) of compositional bias.

Results revealed incorrect grouping of taxa with large compositional bias. Compositional bias couldn’t be the only factor, however, as the fruit fly is biologically more closely related to the honeybee than is the nematode and is slightly more AT-rich than the nematode. Foster et al speculate that this is due to strong amino acid bias—these are elements, which evolved independently and in a similar position in the genome.

In conclusion, Foster et al have shown that compositional bias can take away from the integrity of phylogenetic analysis. With this, we should keep in mind that we should be cautious when interpreting phylogenetic results. It is important to predetermine any compositional biases in your samples.

*^{Foster PG, Hickey DA (1999) Compositional bias may affect both DNA-based and protein-based phylogenetic reconstructions. Journal of Molecular Evolution 48: 283–290. doi: 10.1007/pl00006471.}