*2.3. Mapping and Unification Process*

Once the data were harmonized, a single FCDB was created. The data were differentiated by origin, but organized in a homogeneous structure. The mapping process involved matching foods based on the FoodEx2 identification code. The data were cleaned by eliminating 0 values and treated as missing to eliminate possible errors in the matching. Standard rounding values were taken [43]. Statistical parameters (mean, median, standard deviation) were calculated for each compound whenever a food had the same code. After all the data were evaluated, the coordinators decided to use the median as the final value. Unification was applied to foods with the same codes. The median was used in order to unify and complete the values of a food as long as the matchings were identical. The results were filtered using different filters as values to locate the values of the outer layers. Afterwards, the quality of the data was evaluated. All changes were made in spreadsheets, and Python 3.0 was used for unification and statistical calculations. The scripts used are shown in Supplementary Material S2. For the S4H FCDB, energy was recalculated using the Atwater factors [62]. Once the values were obtained, they could be inputted in the national FCDB for those foods that are not yet included, or for those nutrients or compounds that were missing.
