Detecting selection needs comparative data

Abstract
Arising from: J. B. Plotkin, J. Dushoff & H. B. Fraser Nature 428, 942–945 (2004); see also communication from Hahn et al.; Chen et al.; Plotkin et al. reply Positive selection at the molecular level is usually indicated by an increase in the ratio of non-synonymous to synonymous substitutions (dN/dS) in comparative data. However, Plotkin et al.1 describe a new method for detecting positive selection based on a single nucleotide sequence. We show here that this method is particularly sensitive to assumptions regarding the underlying mutational processes and does not provide a reliable way to identify positive selection.