Incoprea Research — CPIC API field-coverage audit, version 2, 30 October 2026

Working research dataset, not peer reviewed. Not medical advice.
Research/analysis supplied by DeepSeek Harness. This release contains processed,
public aggregate tables. It does not include patient records, raw third-party
snapshots, the unrevised subpages, or the superseded manuscript exports.

Primary source: https://api.cpicpgx.org/v1/
Data model: https://github.com/cpicpgx/cpic-data/wiki
CPIC curated-content license: https://github.com/cpicpgx/cpic-data/blob/main/LICENSE.md
Source snapshot date reported by the supplied analysis: 30 October 2026.

Read with these definitions:
- Empty phenotype-frequency fields are not evidence that research is absent or
  a clinical guideline is unsupported.
- high_risk_prevalence is the supplied column name for the composite proportion
  in CPIC phenotype categories flagged for clinical attention. The article calls
  this the CPIC high-priority proportion. It is not a probability of clinical harm.
- support_n is maximum subjectcount among allele-level records for a gene/group.
  It is a sample-support proxy, not an established phenotype denominator, unique
  person count, or effective sample size.
- Frequency/proportion values are fractions unless a column explicitly indicates
  percentages; ancestry-group counts and pair counts are counts. Missing values
  must not be interpreted as zero. See column names and published definitions.
- coverage_tier grades API field availability, not clinical evidence quality.
- Pair-level results reuse gene-level estimates and are not independent contrasts.
- HLA carrier estimates use Hardy–Weinberg, 1-(1-p)^2.

The processed-table counts were checked at publication. Raw API extraction,
cohort derivations and source-reported phenotype-frequency sum checks were not
independently reproduced by the publisher. No independent clinical review or
external population-genomic replication is claimed.

Published article: https://incoprea.com/research/pharmacogenomics-coverage/
