Figure 4

Partial regression leverage plots from a stepwise regression analysis detailing the contribution of three significant predictors to standardized nucleotide asymmetry between paralogs comprising 130 C. elegans duplicate pairs. The vertical axis represents the partial residuals of standardized nucleotide asymmetry (y variable) and the horizontal axis represents the partial residual of the specific predictor variable. Dashed red lines represent 95% confidence curves for the full model (solid red line) based on actual data. The dashed blue horizontal line indicates a model with no effect. Confidence curves enclosing a solid red line deviating significantly from the dashed blue horizontal line indicate a significant effect of the predictor variable. Alternatively, when the confidence curves enclose the dashed blue horizontal line, the effect of the predictor variable is nonsignificant. (a) Partial regression leverage plot showing a significant positive effect of evolutionary age of duplicates (K _S ) on nucleotide rate asymmetry between paralogs. (b) Partial regression leverage plot showing a significant positive effect of chromosomal location on nucleotide rate asymmetry between paralogs. Rate asymmetry between paralogs increases with relocation of one copy to a different chromosome. (c) Partial regression leverage plot showing a significant negative effect of duplication span on nucleotide rate asymmetry between paralogs.