Here, αtr is a function of the gene’s mean normalized count, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document} $$\bar{\mu}_{i} = \frac{1}{m} \sum\limits_{j} \frac{K_{ij}}{s_{ij}}. $$ \end{document}μ¯i=1m∑jKijsij. It describes the mean-dependent expectation of the prior. σd is the width of the prior, a hyperparameter describing how much the individual genes’ true dispersions scatter around the trend. For the trend function, we use the same parametrization as we used for DEXSeq [30], namely, (6)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document} $$ \alpha_{\text{tr}}(\bar{\mu}) = \frac{a_{1}}{\bar{\mu}} + \alpha_{0}. $$ \end{document}αtr(μ¯)=a1μ¯+α0.
