Fig. 4: Differential allelic dynamics between parents and F1 offspring reveal context-dependent trans-effects in spermatogenesis.
From: The dynamic genetic determinants of increased transcriptional divergence in spermatids
a Interpolated trajectories of allelic imbalance in F0 parental mice (black) and F1 (grey) offspring for Dnajc2. Dots denote average allelic ratios from cells distributed in 100 evenly spaced bins across the differentiation trajectory. b Graphical (top) and mathematical (bottom) representation of the analysis strategy to identify different types of trans-effects. Latent allelic trajectories are modelled by two non-linear functions derived from Gaussian process regression (Methods). Persistent trans-effects manifest as a persistent difference between F0 and F1 allelic ratios (red), dynamic trans-effects as a dynamic difference (orange). Evidence for either type of effect is evaluated using Bayes Factors (BF). c Scatterplot of Bayes Factors that correspond to persistent versus dynamic trans-effects. Horizontal and vertical lines correspond to a logBFs > 10 threshold; the number of genes with evidence for persistent and/or dynamic effects are highlighted. d Box plots of exonic PhastCons scores for genes with no (n = 3227 genes), persistent (n = 313) and dynamic (n = 117) trans-effects (Two-sided Wilcoxon’s rank sum test, ****=6.6e-09, **=0.0014). The boxplots show median, 25%- and 75%-quantiles, the whiskers 1.5 inter-quartile ranges. e Comparison of the fraction of dynamic versus persistent cis- and trans-effects (log Bayes Factor cutoff = 10, set to match the number of dynamic cis effects identified using scDALI; Methods). p = 0.025 using a two-sided Chi-square test. f Scatterplot between log Bayes Factors for persistent and cis- versus persistent trans-effects. Odds ratio of co-occurrence of both effects for the same genes based on logBF threshold 10. g Scatter plot as in (f), however for dynamic cis- versus dynamic trans-effects.
