SNP-HWE also recognizes that it is unnecessary to start the computation with an accurate absolute likelihood for one table. Since the final p-value is computed as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document} $${\fontsize{8.5}{6}\begin{aligned} \frac{[\text{sum of null hypothesis likelihoods of at-least-as-extreme tables}]} {[\text{sum of null hypothesis likelihoods of all tables}]}, \end{aligned}} $$ \end{document}[sum of null hypothesis likelihoods of at-least-as-extreme tables][sum of null hypothesis likelihoods of all tables], it is fine for all computed likelihoods to be relative values off by a shared constant factor, since that constant factor will cancel out. This eliminates the need for log-gamma approximation.
