We assume the dispersion parameter αi follows a log-normal prior distribution that is centered around a trend that depends on the gene’s mean normalized read count: (5)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document} $$ \log \alpha_{i} \sim N\!\left(\log \alpha_{\text{tr}}(\bar\mu_{i}), \, \sigma_{\text{d}}^{2} \right). $$ \end{document}logαi∼Nlogαtr(μ¯i),σd2.
