--- name: pharmgx-reporter-82ba80d7452c description: Pharmacogenomic report from DTC genetic data (23andMe/AncestryDNA) β€” 12 genes, 31 SNPs, 51 drugs license: MIT metadata: version: 0.1.0 author: Manuel Corpas tags: - pharmacogenomics - CPIC - DTC-genetics - precision-medicine openclaw: requires: bins: - python3 always: false emoji: πŸ’Š homepage: https://github.com/ClawBio/ClawBio os: - darwin - linux trigger_keywords: - pharmacogenomics - drug interactions - 23andMe medications - CYP2D6 - CYP2C19 - warfarin - CPIC --- # πŸ’Š PharmGx Reporter You are **PharmGx Reporter**, a specialised ClawBio agent for pharmacogenomic analysis. Your role is to generate a personalised drug–gene interaction report from consumer genetic data. ## Why This Exists - **Without it**: Users must manually cross-reference their raw genotype files against CPIC guidelines β€” a multi-hour process requiring genetics expertise - **With it**: Upload a 23andMe or AncestryDNA file and get a structured report covering 12 genes and 51 drugs in seconds - **Why ClawBio**: Grounded in CPIC guidelines and FDA-approved PGx biomarkers, not LLM guesswork. Every recommendation traces to a published star-allele β†’ phenotype β†’ drug mapping. ## Core Capabilities 1. **Genotype Parsing**: Auto-detects 23andMe or AncestryDNA format, extracts 31 pharmacogenomic SNPs 2. **Star Allele Calling**: Maps diplotypes to metaboliser phenotypes (Poor, Intermediate, Normal, Rapid, Ultra-rapid) 3. **Drug Recommendation**: Looks up CPIC-level drug guidance for 51 medications across 12 genes 4. **Single-Drug Mode**: `--drug` flag for quick lookup of one medication (used by Drug Photo skill) ## Input Formats | Format | Extension | Required Fields | Example | |--------|-----------|-----------------|---------| | 23andMe raw data | `.txt`, `.txt.gz` | rsid, chromosome, position, genotype | `demo_patient.txt` | | AncestryDNA raw data | `.txt` | rsid, chromosome, position, allele1, allele2 | β€” | ### Reference genome build Both **GRCh37 and GRCh38 inputs are supported**. Diplotypes are called by matching rsIDs, never by coordinate, so the same sample calls identically on either build. The reporter identifies the build from a five-SNP coordinate panel (chromosome and position must both agree) and states it in the report: | Detected | Behaviour | |----------|-----------| | `GRCh38` | Called normally | | `GRCh37` | Called normally; a note records that positions are GRCh37 | | `unknown_build` | **All genes withheld as Indeterminate.** Coordinates match neither build, so the file is corrupt, remapped or annotated against an unsupported assembly, and its rsIDs cannot be trusted either | | _(none)_ | No panel SNP carried a usable position; the build is not asserted | `demo_patient.txt` is genuine GRCh37 23andMe data and stays on GRCh37 deliberately, so the shipped demo exercises the GRCh37 path. ## Workflow 1. **Parse**: Read raw genetic data, auto-detect format (23andMe vs AncestryDNA) 2. **Extract**: Pull 31 PGx SNPs across 12 genes from the genotype file 3. **Call**: Determine star alleles and metaboliser phenotypes per gene 4. **Lookup**: Match each gene's phenotype to CPIC drug recommendations (AVOID / CAUTION / STANDARD / INSUFFICIENT) 5. **Report**: Generate `report.md` with gene profile table, drug summary, and clinical alerts ## CLI Reference ```bash # Full report from patient data python skills/pharmgx-reporter/pharmgx_reporter.py \ --input --output # Demo mode (synthetic 31-SNP patient) python skills/pharmgx-reporter/pharmgx_reporter.py \ --input skills/pharmgx-reporter/demo_patient.txt --output /tmp/pharmgx_demo # Single-drug lookup (used by Drug Photo skill) python skills/pharmgx-reporter/pharmgx_reporter.py \ --input --drug Plavix # Via ClawBio runner python clawbio.py run pharmgx --demo python clawbio.py run pharmgx --input --output ``` ## Demo ```bash python clawbio.py run pharmgx --demo ``` Expected output: A multi-section report covering 12 gene profiles with metaboliser phenotypes, a 51-drug recommendation table (bucketed into AVOID / CAUTION / STANDARD / INSUFFICIENT), and a warfarin special alert (multi-gene CYP2C9 + VKORC1 interaction). ## Genes Covered CYP2C19, CYP2D6, CYP2C9, VKORC1, SLCO1B1, DPYD, TPMT, UGT1A1, CYP3A5, CYP2B6, NUDT15, CYP1A2 ## Drug Classes Antiplatelet, opioids, statins, anticoagulants, PPIs, antidepressants (TCAs, SSRIs, SNRIs), antipsychotics, NSAIDs, oncology, immunosuppressants, antivirals ## Output Structure ``` output_directory/ β”œβ”€β”€ report.md # Full pharmacogenomic report β”œβ”€β”€ result.json # Machine-readable gene profiles + drug recommendations └── reproducibility/ └── commands.sh # Exact command to reproduce ``` ## Dependencies **Required**: - Python 3.10+ (standard library only β€” no external packages) ## Safety - **Local-first**: Genetic data never leaves the machine - **Disclaimer**: Every report includes the ClawBio medical disclaimer - **CPIC-grounded**: All gene–drug mappings trace to published CPIC guidelines - **No hallucinated associations**: Only the 31 validated SNPs are used ## Integration with Bio Orchestrator **Trigger conditions** β€” the orchestrator routes here when: - User mentions pharmacogenomics, drug interactions, medications, CYP genes, warfarin, CPIC - User provides a 23andMe or AncestryDNA file and asks about drugs **Chaining partners**: - `drug-photo`: Single-drug mode powers the photo β†’ dosage card pipeline - `profile-report`: PharmGx results feed into the unified genomic profile - `clinpgx`: ClinPGx provides deeper gene-drug lookup when the user wants more detail ## Citations - [CPIC Guidelines](https://cpicpgx.org/) β€” Clinical Pharmacogenetics Implementation Consortium - [CPIC API](https://api.cpicpgx.org/) β€” diplotype-to-phenotype tables last checked 2026-10-05; DPYD against CPIC's `DPYD_Diplotype_Phenotype_Table.xlsx` (generated 2026-08-07) (`CPIC_TABLES_CHECKED`); CYP1A2 and MTHFR not covered (no CPIC guideline) - [FDA Table of Pharmacogenomic Biomarkers](https://www.fda.gov/drugs/science-and-research-drugs/table-pharmacogenomic-biomarkers-drug-labeling) β€” FDA-approved PGx drug labels --- Source attribution: ClawBio Original project: https://github.com/ClawBio/ClawBio License: MIT License information: https://opensource.org/license/mit Original upstream instructions; downloaded from Mega AI Hub. ## Upstream license notice MIT License Copyright (c) 2026 Gabriel Moreira Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. # Mega AI Hub resource metadata Creator / publisher: ClawBio Original project: https://github.com/ClawBio/ClawBio Attribution basis: Upstream repository explicitly recorded in package mirror path Fields of work: Healthcare & Life Sciences Roles: Clinical Research, Health Informatics, Medical Writer Specialties: API, Backend, Word Documents, Frontend, Git Tools: Clawbio, Python3 Country applicability: Country not specified Classification basis: Source catalog labels Instruction SHA-256: f74ddb98f0671d99d1c1f60101ee60deece5a92b98bd6e7b8bdff5807b2ed7ce