New mart: Products to inactive ingredients (excipients)

[jrlegrand]

Resolves #296 Resolves #276

Explanation

• Create a products to inactive ingredients mart using data from DailyMed via RxNorm (SAB = MTHSPL). This links MTHSPL products (TTY = DP) with MTHSPL substances (TTY = SU) as well as associating with a 9-digit NDC (i.e. the first two segments of the NDC which represent the product, but not the packaging, etc).
• Added a DAG to automate pulling in UNII Code data from FDA.
• Added some logic to try to pull in UNII codes by RXCUI where a UNII code did not exist.
• Joined UNII standard normalized display names to the products to inactive ingredients mart.
• Cleaned up duplicate product names and excipient names by using string_agg() and pipe-delimiting (' | '). OTC products were the larger offenders here for product names (i.e. lip balms with dozens of flavors that are all essentially the same product). As far as I can tell, sometimes rxnorm_rxnrel would map to multiple of the same excipient by RXAUI which are essentially the same inactive ingredient by UNII / RXCUI. The solution was to just string_agg the ingredient names in these situation as we are primarily going to be using the UNII standard display name.
• Added a seed file that looks to USP preservative products by CAS RN identifier (based on the 29 results at the time of writing via this catalog page: https://store.usp.org/preservatives/category/USP-1213).
• Used the above seed file in conjunction with the FDA UNII file (which has a UNII -> CAS RN mapping) to flag UNIIs in the data mart as being considered preservatives by USP.
• Included the PubChem ID from the FDA UNII file. You can create a link to a PubChem detail page by concatenating the URL with the ID. Example: concat('https://pubchem.ncbi.nlm.nih.gov/compound/', pubchem_id)

Rationale

Removed duplicates for data quality / granularity.

Added preservative flag to facilitate people searching for products that contain preservatives. Can SUM this column in aggregate to identify products that contain any preservatives.

Names of excipients are not normalized in RxNorm. Oftentimes, it is just a matter of different capitalization that can be solved by lowercasing everything, but other times there are extra words or formatting differences. Normalizing the names is a better user experience and potentially better for reporting.

Tests

Ran mart and explored data in pgadmin.

[jrlegrand]

See attached file for excipients flagged as preservatives (per the USP methodology) that are in the mart at least once, with counts for how frequently they appear. Also has a link to PubChem to demonstrate this functionality and make it easier to review. Note: one missing PubChem link due to how the data exists from FDA UNII file.

preservatives_for_review.xlsx

[jrlegrand]