Table 1.
Parameter names, description, design values, and priors for Bayesian inference.
Table 2.
Parameter names, description, and priors for Gaussian Process Bayesian inference.
Fig 1.
Response to artificial selection for night sleep.
(A) Mean and (B) coefficient of environmental variation of night sleep. Plot and regression lines of cumulated selection differential (ΣS) against cumulated selection response (ΣR) for (C) long- and (D) short-sleeping populations, and against cumulated differential ΣD for (E) controls. Light green, Replicate 1 long-sleeper population; Dark green, Replicate 2 long-sleeper population; Orange, Replicate 1 short-sleeper population; Red, Replicate 2 short-sleeper population; Gray, Replicate 1 control population; Black, Replicate 2 control population.
Fig 2.
Fit of Hierarchical Generalized Linear Model to gene CG1304 for flies selected for short sleep, unselected controls, and selected for long sleep.
The solid lines show the expected value of full model, dashed lines for reduced model, and shaded regions show the 95% credibility interval. Replicate 1 data points are shown in dark gray, Replicate 2 in light gray.
Fig 3.
Fit of Gaussian Process model to pair of genes LysC and CG1304, for female flies selected for short sleep, unselected controls, and selected for long sleep.
The solid lines show the expected value, while the shaded regions show the 95% credibility interval. Replicate 1 data points are shown in dark gray, Replicate 2 in light gray). The expectation for correlations (ρsel) is shown for each selection scheme. An asterisk indicates significant difference from controls in selection scheme, as opposed to non-significance (n.s.).
Fig 4.
Signal variances and covariances normalized to range [-1,1] for females and males in each of the selection schemes: Short, control, and long.
Each off-diagonal square is the expected value of the interaction between two of 85 genes, for a total of 3,570 pairs.
Fig 5.
Gene interaction network in males selected for long sleep.
Edges represent signal covariances whose posterior distributions do not overlap with that of controls at 95% credibility. Colors and line thickness indicate indicate the strength and the direction of the correlation. Thin gray lines show all 145 interactions significant for at least one of the four sex-selection scheme combinations.
Fig 6.
Night sleep in Minos mutations.
The figure compares night sleep duration in each mutant with that of its respective control. Blue circles indicate Minos insertions with w1118 control strain; red triangles indicate Minos insertions with y1w67c23 control strain. **** or ####, P < 0.0001; ***, P < 0.001, three-way ANOVA. All P-values are less than the Bonferroni-corrected P-value of 0.0125.
Fig 7.
Comparison of ratios of gene expression between genes with significant Gaussian Process correlations and unrelated genes for CG12560 and Jon65Aii mutants.
Purple lines show the ratio of mutant gene expression to control for genes with significant Gaussian Process correlations. The distribution of gene ratios for 1, 000 unrelated genes is plotted in the background. Genes having the most extreme ratios are indicated; see S14 Table for the calculations. (A) CG12560 females; (B) CG12560 males; (C) Jon65Aii females; (D) Jon65Aii males.