Analyze Genome Data from VCF Files

Genome Analyzer turns a raw VCF genome file into a plain-language answer to a specific genetics question - about hereditary traits, disease predisposition, drug/substance metabolism (caffeine, alcohol, medication), sports ability, or nutrition. It works in three steps: it locates the VCF file in the current directory with Glob (pattern *.vcf), spins up a general-purpose research subagent (via the Task tool, model sonnet) that searches the GWAS Catalog, SNPedia, and scientific literature for SNPs (rsIDs) relevant to the question, then Greps the VCF file for those specific rsIDs and interprets the genotype found at each one.

Use it when a user asks about their own genetics - hereditary traits, disease risk, substance metabolism, fitness predisposition, or gene-based nutrition - and has a VCF file available in the working directory. It is explicitly informational, not diagnostic: the skill's own report template ends with a disclaimer that "genetics is probability, not destiny," that environment and lifestyle often matter more than genes, and that the analysis is not medical advice. Do not use it as a substitute for a clinical genetic counselor or a doctor, and do not use it if no VCF file is present - the skill asks the user which file to use if several are found, or reports that none exists.

Input is a single VCF file (any filename, *.vcf) in the current directory, plus the user's natural-language question about a genetics topic. Internally, genotypes are read directly from the VCF's calls: 0/0 means homozygous for the reference allele, 0/1 heterozygous, and 1/1 homozygous for the alternate allele; each is matched against the risk/protective interpretation of that allele gathered in the research step. Output is a single structured Markdown report following a fixed template: a table of SNP/gene/genotype/risk-level/interpretation, a "what this means" section covering strengths and areas of attention, a numbered list of practical recommendations, a summary risk-rating table, and a disclaimer with source links to SNPedia and the GWAS Catalog. The report is deliberately visual, using emoji risk markers (protective/neutral/moderate/high) and topic icons (sports, weight, caffeine, alcohol, nutrition, medication, cardiovascular, general genetics).

The skill composes two other Claude Code primitives rather than calling an external API directly: it dispatches a general-purpose subagent via the Task tool (model sonnet) to do the actual SNP research using WebSearch against GWAS Catalog and SNPedia, then uses Glob and Grep locally to find the VCF file and pull out the relevant lines. There is no bundled genome database or bioinformatics library - all genetic reference data comes from the live web research step.

Individuals who have their own VCF genome file (e.g. from a consumer DNA testing service) and want a structured, sourced, plain-language walkthrough of what specific genes/SNPs say about a topic they're curious about - rather than raw variant calls or a wall of clinical jargon.