Comment on "mbtransfer: Microbiome intervention analysis using transfer functions and mirror statistics": Implementation errors, theoretical misapplication, and methodological flaws

Abstract: There are a number of errors in "mbtransfer: Microbiome intervention analysis using transfer functions and mirror statistics" PLOS Comp Bio (2024) spanning multiple aspects of the paper. The wrong inputs were provided to comparator methods for model training, when forecasting one method was provided initial conditions in the wrong units, and performance metrics were calculated without proper unit conversion. The false discovery rate and power analysis conclusions provided in the text are not supported by theory or the empirical testing that was performed within the paper. The paper also has data leakage issues, equations are written down incorrectly, and incorrect definitions/terminology are used.