Figure 1.

Read assignment. Reads aligning to the maternal allele (m) are purple, those aligning to the paternal allele (p) are green, and those aligning equally well to both alleles (unassigned) are gray. Expected values according to the environmental model are given in Table 1, and are in black type above. The baseline model uses information of the reads aligning to paternal or maternal alleles only (red boxes), while the environmental model additionally incorporates information about unassigned reads (blue boxes). In the baseline model, $\mu$ is the overall expected value, ${\beta }_k$ models the biological replicate variation, $\alpha \ne 1$ is equivalent to AI, and the known quantity $1-q$ is interpreted as the expected proportion of maternal read counts when there is no AI. In other terms, $E\left(x/\left(y+x\right)\mathrm{|}\alpha =1\right)=1-q.$ The priors for all model parameters are gamma(1/2,1/2). The quantities $\sqrt{\alpha }\mu {\beta }_k$ and $1/\sqrt{\alpha }$ in the notation of the baseline model play the role of the parameters ${\beta }_k/\left(r_m+r_p\right)$ and $\alpha$ respectively in the environmental model. When $\alpha =1$ in the environmental model, $E\left(x\mathrm{|}\alpha =1\right)/E\left(x+y\mathrm{|}\alpha =1\right)=r_m/\left(r_m+r_p\right).$ This leads us to an interpretation of $q$ under the null in the baseline model as $r_p/\left(r_m+r_p\right)$ in the environmental model.