Hi @nathandelang ,
See chapter 10 of the Tarquin User Guide; the internal water scaling is fairly rudimentary, basically relying on a supplied (or default) attenuation factor and concentration for water. This is similar to the approach used in LCModel.
To account for different composition, one approach may be to calculate a combined scaling factor (eg, per Gasparovic et al 2006, Near et al 2020) and supply that as suggested in the absolute quantification thread you mentioned. However, this won’t allow for metabolite-specific relaxation parameters to be considered. Another possibility is to quantify initially with the default scaling, then numerically reverse this before applying per-subject, metabolite-specific scaling.