ncbi_clinvar

ClinVar Database (NCBI)

Overview

ClinVar is NCBI's archive of relationships between human genetic variants and phenotypes, with supporting evidence. In this project the agent exposes one compound download tool (download_ncbi_clinvar_variants) that searches by term, retrieves variant IDs, and saves results to JSON. For finer control, ncbi_operations.py also provides atomic ClinVar E-utilities download/query functions (esearch, esummary, efetch), and the low-level module ncbi_clinvar.py provides term builders and FTP helpers. All download/query functions return rich JSON {status, file_info/content, content_preview, biological_metadata, execution_context}.

When to Use This Skill

• Searching for variants by gene, condition, or clinical significance
• Interpreting clinical significance (pathogenic, benign, VUS); see references/clinical_significance.md
• Accessing ClinVar via E-utilities (esearch, esummary, efetch) or FTP
• Understanding review status (star ratings) and conflicting interpretations
• Annotating variant call sets with clinical significance; see references/data_formats.md for VCF/XML/tab formats

Quick Start

The skill provides:

1. Project modules in src/tools/database/ncbi/: ncbi_clinvar.py (atomic E-utilities, term builder, FTP), ncbi_operations.py (query/download operations); ClinVar download functions re-exported via package.
2. References: references/api_reference.md, references/clinical_significance.md, references/data_formats.md

Agent Tools (Download Only)

Also available as general NCBI tools (not ClinVar-specific but useful in ClinVar workflows):

Project Modules (Programmatic Use)

Core Capabilities

1. Search ClinVar Variants (download)

from src.tools.database.ncbi import download_ncbi_clinvar_variants

# Simple gene search
result = download_ncbi_clinvar_variants("BRCA1[gene]", "output/clinvar_brca1.json", retmax=50)

# Pathogenic variants for a gene
result = download_ncbi_clinvar_variants(
    "BRCA1[gene] AND pathogenic[CLNSIG]",
    "output/clinvar_brca1_pathogenic.json",
    retmax=100,
)

2. Build Search Terms (programmatic)

from src.tools.database.ncbi.ncbi_clinvar import build_clinvar_term

# Build term with filters
term = build_clinvar_term(
    gene="BRCA1",
    clinical_significance="pathogenic",
    exclude_conflicting=True,
)
# → "BRCA1[gene] AND pathogenic[CLNSIG] NOT conflicting[RVSTAT]"

# Search by condition
term = build_clinvar_term(condition="breast cancer")
# → "breast cancer[disorder]"

3. Atomic E-utilities Workflow (programmatic)

from src.tools.database.ncbi.ncbi_clinvar import (
    build_clinvar_term, query_clinvar, parse_esearch_ids,
    get_clinvar_summary, fetch_clinvar_records,
)

# Step 1: Build term and search
term = build_clinvar_term(gene="BRCA1", clinical_significance="pathogenic")
esearch_result = query_clinvar(term, retmax=100)

# Step 2: Parse IDs
variation_ids = parse_esearch_ids(esearch_result)

# Step 3: Get summaries or full records
summaries = get_clinvar_summary(variation_ids)
records = fetch_clinvar_records(variation_ids, rettype="vcv")

4. Download Atomic Results

from src.tools.database.ncbi import (
    download_ncbi_clinvar_esearch,
    download_ncbi_clinvar_esummary,
    download_ncbi_clinvar_efetch,
)

# Save esearch results
download_ncbi_clinvar_esearch("BRCA1[gene]", "output/clinvar_esearch.json", retmax=50)

# Save summaries (need IDs from esearch first)
download_ncbi_clinvar_esummary(["12345", "67890"], "output/clinvar_esummary.json")

# Save full records
download_ncbi_clinvar_efetch(["12345", "67890"], "output/clinvar_efetch.xml")

5. FTP Bulk Downloads (programmatic)

from src.tools.database.ncbi.ncbi_clinvar import get_clinvar_ftp_url, download_clinvar_ftp

# Get FTP URL
url = get_clinvar_ftp_url("variant_summary")  # Or: vcf_grch37, vcf_grch38, xml_latest, var_citations, cross_references

# Download
download_clinvar_ftp("variant_summary", "output/clinvar_ftp/")
download_clinvar_ftp("vcf_grch38", "output/clinvar_ftp/")

Key Concepts

• Search fields: [gene], [CLNSIG], [disorder], [variant name], [chr], [RVSTAT] (e.g. exclude conflicts: NOT conflicting[RVSTAT])
• Variation ID (VCV): ClinVar variation accession; returned by esearch as UID and used in esummary/efetch
• rettype: vcv (variant-centric) or rcv (variant–condition pair) for efetch

Important Notes

• ClinVar data is not for direct clinical diagnosis; involve a genetics professional
• Prefer high review status (e.g. expert panel, practice guideline) when interpreting classifications
• Use exclude_conflicting=True in build_clinvar_term to filter out conflicting interpretations
• Rate limits: 3 req/s without API key, 10 with key; the module applies delays automatically

Reference Documentation

• references/api_reference.md – E-utilities (esearch, esummary, efetch, elink), parameters, rate limits, Biopython example
• references/clinical_significance.md – Germline/somatic classifications, ACMG/AMP terms, review status, conflict resolution
• references/data_formats.md – FTP layout, XML (VCV/RCV), VCF, tab-delimited files, processing examples