Consider incorporating UNII database into build

[hoganwr]

RxNorm includes Unique Ingredient Identifiers (UNIIs). So once we figure out how to handle DrugBank identifiers, we can use the same process to include UNIIs. Then we can bring in some nice data/info from the UNII database. For one thing it has a view to classifying certain ingredients as biologics.

[hoganwr]

Verified that ChEBI is not importing UNIIs. Looks like maybe 2 total (for all of ChEBI not just the import into DrOn) and they seemed to be there by accident.

[hoganwr]

Also there will be a challenge getting them out of RxNorm, because they are mixed in biologics codes:

U.S. Food and Drug Administration (FDA) generated unique ingredient identifiers (UNIIs) are available in RXNCONSO.RRF as the 'CODE' value for SAB=MTHSPL TTY=SU atoms. These UNII codes are pulled directly from the ingredient information included in the structured product labels (SPLs) on DailyMed .

The UNII is a non-proprietary, free, unique, unambiguous, non semantic, alphanumeric identifier based on a substance’s molecular structure and/or descriptive information. For more information on the FDA UNII codes, please refer to this FDA web page .

A portion of SAB=MTHSPL TTY=SU atoms do not have UNIIs as the 'CODE' value. Biological product names impacted by FDA's guidance on Nonproprietary Naming of Biological Products contain a biological drug substance code instead of a UNII in the 'CODE' field. Biological drug substance codes are obtained from the SPL Biologic or Drug Substance Indexing File available on DailyMed . To learn more about biological product names, refer to FDA’s Nonproprietary Naming of Biological Products .

So we'd have to compare with UNII database codes and the SPL Biologic ... File. Complicates things.

[hoganwr]

This can be done entirely from UNII database side, as UNII provides RxCuis for many ingredients. Of 167,683 UNII database entries, approximately 13,708 have an RxCUI (as of the Feb 11, 2026 file).

Note too that it is possible to download archive UNII files dating back to 2014/2015. Need to think about whether that is worthwhile at all (for DrugBank so far, we just decided to get them going forward for now -- a big historical capture is a significant project).