| name | genetic-testing-report |
| description | Generate clinical single-gene genetic disease testing reports compliant with T/SZGIA 4-2018. Use when the user asks for monogenic genetic test reports, variant interpretation reports, 单基因遗传病检测报告, 基因检测报告规范, T/SZGIA 4-2018, or structured PGx/clinical genetics report output. |
Clinical Single-Gene Genetic Testing Report (T/SZGIA 4-2018)
Overview
Generate reports for clinical single-gene genetic disease testing following
group standard T/SZGIA 4-2018 (临床单基因遗传病基因检测报告规范).
| Standard | T/SZGIA 4-2018 |
|---|
| Scope | Reports from third-party labs, hospital labs, university diagnostic labs |
| Structure | Main body (essential findings) + Appendix (supplementary detail) |
Do not issue a disease diagnosis in the report. State findings and
interpretation; recommend genetic counseling and confirmatory testing when
appropriate.
When to apply
- User requests a monogenic / single-gene genetic test report
- User provides gene, variant(s), and clinical indication for report drafting
- User asks to format output per T/SZGIA 4-2018 or Chinese PGx report norms
- Agent tool
generate_genetic_testing_report is invoked
Mandatory report sections (13 items)
Every complete report must include all 13 sections below. Missing sections
are non-compliant.
| # | Section | Required in main body |
|---|
| 1 | Testing institution info & contact | Yes |
| 2 | Subject (patient) basic info | Yes |
| 3 | Referring institution & physician | Yes |
| 4 | Sample type | Yes |
| 5 | Sample receipt date & report date | Yes |
| 6 | Test item & method | Yes (summary); details in appendix |
| 7 | Test results (variants) | Yes (table) |
| 8 | Test conclusion | Yes |
| 9 | Result interpretation | Yes |
| 10 | Recommendations | Yes |
| 11 | References | Yes |
| 12 | Method description & limitations | Appendix preferred |
| 13 | Signatures & seal | Yes (placeholders if draft) |
Main body vs appendix
Main body — concise, clinician-facing:
- Subject & test metadata (sections 1–6, abbreviated)
- Primary results: variants related to the clinical indication
- Secondary / incidental findings (only if consented and clinically actionable)
- Conclusion, interpretation, recommendations
Appendix — supplementary:
- Full gene panel / exome scope
- Disease background for detected genes
- Detailed method (library prep, sequencing platform, bioinformatics pipeline)
- QC metrics (coverage, depth, sensitivity/specificity)
- Benign / VUS variants not reported in main body (optional listing)
- ACMG evidence details per variant
Section content requirements
1–6. Test metadata
Subject info must include (when available):
- Name, sex, date of birth
- Test date / sample accession date
- Test purpose & clinical indication: symptoms, age of onset, imaging/pathology,
provisional diagnosis, family history, consanguinity
Sample types: peripheral blood, DNA, dried blood spot, saliva, tissue, etc.
Test method summary: e.g. targeted NGS panel, WES, Sanger confirmation.
7. Test results
For each reported variant, include:
| Field | Requirement |
|---|
| Gene | HGNC symbol |
| Transcript | RefSeq/Ensembl ID with version; state canonical transcript rationale |
| Variant | HGVS c. / p. / g. nomenclature |
| Genotype | het / hom / hemizygous / mosaic (with fraction if known) |
| Zygosity context | de novo / inherited / unknown |
| Allele frequency | gnomAD, ExAC, 1000G, internal cohort |
| In silico prediction | SIFT, PolyPhen, CADD, splice predictors (if applicable) |
| ClinVar / HGMD | Known classification & ID |
| ACMG classification | Pathogenic / Likely pathogenic / VUS / Likely benign / Benign |
| Evidence codes | PM2, PP3, PS1, etc. (when classified) |
Present primary variants in a table in the main body.
8. Test conclusion
Reporting rules (ACMG-aligned):
| Variant class | Report in conclusion? |
|---|
| Pathogenic / Likely pathogenic, related to indication | Yes |
| VUS, related to indication | Only when conclusion is pending / unresolved |
| Benign / Likely benign | No |
| Secondary / incidental findings | Only if consented; clinically actionable only |
| Unrelated VUS | No |
Use clear language: "detected", "consistent with", "cannot exclude" — avoid
definitive disease diagnosis.
9. Result interpretation
For each reported variant:
- ACMG evidence summary (database hits, literature, segregation, de novo status)
- Phenotype correlation with subject's clinical presentation
- Inheritance pattern (AD / AR / XL / mitochondrial)
- Recurrence risk for family members
10. Recommendations
Typical items:
- Genetic counseling referral
- Familial segregation / parental testing
- Expanded testing if negative or inconclusive
- Specialist referral aligned with phenotype
- Re-analysis when new evidence emerges
11. References
Vancouver or numbered style. Include PMIDs/DOIs for all citations used in
pathogenicity assessment.
12. Method & limitations
State:
- Specimen requirements & extraction method
- Sequencing platform & read length
- Target region / gene list / exome coverage
- Variant calling & annotation pipeline
- Validation method for reportable variants (Sanger, etc.)
- Limitations: regions not covered, genes not tested, types of variants
not detectable (CNV, repeat expansions, methylation, low-level mosaicism)
- Analytical sensitivity & specificity for SNV/indels (and CNV if applicable)
13. Signatures & seal
- Report author (检验者)
- Result reviewer (核对者)
- Final approver (签发者)
- Third-party lab: "检测报告专用章" placeholder
- Report version / revision note if amended
Tool: generate_genetic_testing_report
When exposed as an LLM tool, accept structured input and return JSON matching
report-schema.json.
Input parameters
| Parameter | Type | Required | Description |
|---|
testing_institution | object | Yes | Name, address, phone, lab license |
subject | object | Yes | Patient demographics & clinical indication |
referring | object | Yes | Referring hospital & physician |
sample | object | Yes | Type, collection date, receipt date |
test | object | Yes | Panel name, method, genes covered |
variants | array | Yes | Detected variants (see schema) |
conclusion | string | Yes | Primary conclusion text |
interpretation | string | Yes | Detailed interpretation |
recommendations | array[string] | Yes | Action items |
references | array[object] | Yes | {pmid, doi, citation} |
method_limitations | object | Yes | Method summary + limitation list |
signatures | object | No | Author, reviewer, approver names |
include_appendix | bool | No | Default true |
Output
Return JSON conforming to report-schema.json. Also produce a human-readable
Markdown summary in report_markdown field.
Variant classification quick reference
Follow ACMG/AMP 2015 (PMID: 25741868) and ClinGen updates:
- Combine evidence using PS / PM / PP / BA / BS / BP codes
- Report only classifications with sufficient evidence
- Never upgrade VUS to pathogenic without strong evidence
- Document every evidence item used
Quality checklist
Before finalizing, verify:
Report compliance:
- [ ] All 13 T/SZGIA sections present
- [ ] HGVS nomenclature used for all variants
- [ ] Transcript ID and version specified
- [ ] ACMG class stated for each reported variant
- [ ] No benign/LB variants in conclusion
- [ ] VUS reporting follows indication-only rule
- [ ] No definitive disease diagnosis language
- [ ] Limitations section covers method scope
- [ ] References match cited evidence
- [ ] Signatures / seal placeholders included
Example conclusion phrasing
Good:
In the CFTR gene, a likely pathogenic variant c.1521_1523del (p.Phe508del)
was detected in the homozygous state, consistent with the clinical suspicion
of cystic fibrosis. Genetic counseling and familial carrier testing are
recommended.
Avoid:
The patient has cystic fibrosis. (diagnosis — not allowed)
Additional resources