Reject Self Reference

Goal: build analysis records for alanine aminotransferase (ALT), aspartate aminotransferase (AST), and their ratio (ASTALT), carrying ADT and AVAL for each subject, date, and parameter.

Input: collected results carrying the collection date, test code (either ALT or AST), and numeric result, plus a file of analysis values.

Variables:

• ADT would be the date the result was collected.
• AVAL would be the collected result on each transaminase record. On the ratio record it would be the aspartate value for the same subject and date, read from the dataset being built, divided by the alanine result; missing when either result is absent or the alanine result is zero.

The aspartate values the ratio reaches for are the values this run is producing under the same name. Nothing can distinguish the record being written from the record being read back, so the run is rejected before any data is read and no artifact is accepted.

How to fix

First decide whether the aspartate values the ratio reads are already final. If they are, they belong to a finished dataset that this run merely reads, so give that source a name of its own and keep the name of the dataset being built for the dataset being built:

input:
  LB: {path: input/lb.csv, types: {LBSTRESN: float}}
  ADLB_RAW: {path: input/adlb.csv, types: {AVAL: float, ADT: date}}

Renaming the declaration alone leaves the read pointing at a name that no longer exists, so change where it reads from as well:

intermediates:
  - id: ADLB_LOOKUP
    dataset: ADLB_RAW

The read then uses a completed dataset by its own name, which is an ordinary source like any other. If the aspartate values are not final, splitting the work applies instead: complete and write the transaminase parameters before the run that reads them starts.