FIG. 2.

(A) Gene lengths between Apis mellifera and Drosophila melanogaster are highly correlated. Ortholog length in D. melanogaster can explain 41% of observed variation in A. mellifera gene lengths in a linear regression model (see text). Note that the lengths are log-transformed to improve normality. (B) Residuals remaining from the regression model in figure 2A. Residuals from low-CpG genes (red triangles) tend to be negative, whereas those from the high-CpG genes (green circles) tend to be positive, demonstrating that low-CpG genes are shorter and high-CpG genes are longer than expected from the linear regression model alone. Note that there exists a subset of high-CpG genes with particularly large residuals (denoted as darker green circles). These genes include those related to specific developmental functions (see text).