Specifying background set of compounds

Please note, that by default, it is supposed that your compounds were selected from the set of all KEGG compounds (compounds that lies on global metabolite transformation network). This can be far from truth in many cases and cause incorrect estimate of the p-values of the models. If your list of compounds was selected from a set which is biased then to get correct estimates of the p-value you should also provide a background set of compounds, e.g. the set of all measured compounds in your experiments. In this case, random sampling would be performed using the supplied background set.