Uncovering the mutation-fixation correlation in short lineages

Background: We recently reported a highly unexpected positive correlation between the fixation probability of nonsynonymous mutations (estimated by omega) and neutral mutation rate (estimated by K(s)) in mammalian lineages. However, this positive correlation was observed for lineages with relatively long divergence time such as the human-mouse lineage, and was not found for very short lineages such as the human-chimpanzee lineage. It was previously unclear how to interpret this discrepancy. It may indicate that the positive correlation between omega and K(s) in long lineages is a false finding. Alternatively, it may reflect a biologically meaningful difference between various lineages. Finally, the lack of positive correlation in short lineages may be the result of methodological artifacts. Results: Here we show that a strong positive correlation can indeed be seen in short lineages when a method was introduced to correct for the inherently high levels of stochastic noise in the use of K(s) as an estimator of neutral mutation rate. Thus, the previously noted lack of positive correlation between omega and K(s) in short lineages is due to stochastic noise in K(s) that makes it a far less reliable estimator of neutral mutation rate in short lineages as compared to long lineages. Conclusion: A positive correlation between omega and K(s) can be observed in all mammalian lineages for which large amounts of sequence data are available, including very short lineages. It confirms the authenticity of this highly unexpected correlation, and argues that the correction likely applies broadly across all mammals and perhaps even non-mammalian species.