[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6DAYTM5A1FES1KZE2HCEEPR":3,"skill-same:01M6DAYTM5A1FES1KZE2HCEEPR":107},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":28,"linkOnly":29,"markdownContent":30,"markdownBody":31,"frontMatter":32,"refName":33,"commitSha":34,"skillDirectory":35,"forksCount":14,"repositoryHtmlUrl":36,"repositoryHomepageUrl":6,"ownerHtmlUrl":37,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":38,"install":39,"resources":52},"01M6DAYTM5A1FES1KZE2HCEEPR","tooluniverse-variant-analysis",null,"Use when parsing VCF files, filtering variants, annotating mutation consequences, or interpreting SNVs, indels, CNVs, and structural variants.","Prekzursil\u002Fagent-skills-toolchain","Prekzursil","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F54636077?v=4","01M6DACZ1PCQHD99SCR1YXG28B","Andrei Mihai Visalon","01M6DACZ670WZF535V3WYYSS4Z",0,"MIT","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Flibrary\u002Fskills\u002Ftooluniverse-variant-analysis\u002FSKILL.md",[],[19,20,21,22,23,24,25,26,27],"analysis","variant analysis","tooluniverse variant analysis","parsing","filtering","use","files","mutation","structural","2026-11-10 10:12:30",false,"---\nname: tooluniverse-variant-analysis\ndescription: Use when parsing VCF files, filtering variants, annotating mutation consequences, or interpreting SNVs, indels, CNVs, and structural variants.\n---\n\n# Variant Analysis and Annotation\n\nProduction-ready VCF processing and variant annotation skill combining local bioinformatics computation with ToolUniverse database integration. Designed to answer bioinformatics analysis questions about VCF data, mutation classification, variant filtering, and clinical annotation.\n\n## Domain Reasoning\n\nVCF quality filtering must come before interpretation. A variant called at 2x read depth is unreliable regardless of its QUAL score, because stochastic sequencing errors at low depth can mimic true variants. The recommended minimums — depth > 10x, QUAL > 20, allele frequency consistent with expected zygosity — are not conservative; they are the floor below which calls cannot be trusted. Applying lenient filters to \"keep more variants\" sacrifices accuracy for coverage and produces false positives that propagate through all downstream analyses.\n\n## LOOK UP DON'T GUESS\n\n- Clinical significance of specific variants: query `MyVariant_query_variants` or `EnsemblVEP_annotate_rsid`; never cite ClinVar classifications from memory.\n- Population allele frequencies: retrieve from MyVariant.info or gnomAD tools; do not assume rarity.\n- ClinGen dosage sensitivity scores for genes in a CNV: call `ClinGen_dosage_by_gene`; do not estimate HI\u002FTS scores.\n- Mutation consequence predictions: run Ensembl VEP or retrieve from MyVariant.info; do not classify impact without tool output.\n\n---\n\n## CRISPR sgRNA Design Reasoning\n\n- PAM sequence (NGG for SpCas9) must lie 3' of the target on the non-target strand; the guide RNA targets the 20 nt immediately upstream of the PAM\n- For exon targeting: choose guides that cut early in the coding sequence for maximum frameshift\u002Fdisruption\n- Off-target risk increases with fewer mismatches; always check for genomic sites with 0-3 mismatches to the guide\n\n---\n\n## When to Use This Skill\n\n**Triggers**:\n- User provides a VCF file (SNV\u002Findel or SV) and asks questions about its contents\n- Questions about variant allele frequency (VAF) filtering\n- Mutation type classification queries (missense, nonsense, synonymous, etc.)\n- Structural variant interpretation requests (deletions, duplications, CNVs)\n- Variant annotation requests (ClinVar, gnomAD, CADD, dbSNP)\n- CNV pathogenicity assessment using ClinGen dosage sensitivity\n- Cohort comparison questions\n- Population frequency filtering (SNVs or SVs)\n- Intronic\u002Fintergenic variant filtering\n- Gene dosage sensitivity queries\n\n**Example Questions**:\n- \"What fraction of variants with VAF \u003C 0.3 are annotated as missense mutations?\"\n- \"After filtering intronic\u002Fintergenic variants, how many non-reference variants remain?\"\n- \"What is the clinical significance of this deletion affecting BRCA1?\"\n- \"Which dosage-sensitive genes overlap this 500kb duplication on chr17?\"\n- \"How many variants have clinical significance annotations?\"\n- \"Compare variant counts between samples\"\n\n---\n\n## Core Capabilities\n\n| Capability | Description |\n|-----------|-------------|\n| **VCF Parsing** | Pure Python + cyvcf2 parsers. VCF 4.x, gzipped, multi-sample, SNV\u002Findel\u002FSV |\n| **Mutation Classification** | Maps SO terms, SnpEff ANN, VEP CSQ, GATK Funcotator to standard types |\n| **VAF Extraction** | Handles AF, AD, AO\u002FRO, NR\u002FNV, INFO AF formats |\n| **Filtering** | VAF, depth, quality, PASS, variant type, mutation type, consequence, chromosome, SV size |\n| **Statistics** | Ti\u002FTv ratio, per-sample VAF\u002Fdepth stats, mutation type distribution, SV size distribution |\n| **Annotation** | MyVariant.info (aggregates ClinVar, dbSNP, gnomAD, CADD, SIFT, PolyPhen) |\n| **SV\u002FCNV Analysis** | gnomAD SV population frequencies, DGVa\u002FdbVar known SVs, ClinGen dosage sensitivity |\n| **Clinical Interpretation** | ACMG\u002FClinGen CNV pathogenicity classification using haploinsufficiency\u002Ftriplosensitivity scores |\n| **DataFrame** | Convert to pandas for advanced analytics |\n| **Reporting** | Markdown reports with tables and statistics, SV clinical reports |\n\n---\n\n## Workflow Overview\n\n**Phase 1: Parse VCF** → Extract CHROM\u002FPOS\u002FREF\u002FALT\u002FQUAL\u002FFILTER\u002FINFO, per-sample GT\u002FVAF\u002Fdepth, annotations (ANN\u002FCSQ\u002FFUNCOTATION). Pure Python or cyvcf2.\n\n**Phase 2: Classify** → Variant type (SNV\u002FINS\u002FDEL\u002FMNV\u002FSV), mutation type (missense\u002Fnonsense\u002Fsynonymous\u002Fframeshift\u002Fsplice\u002Fetc.), impact (HIGH\u002FMODERATE\u002FLOW\u002FMODIFIER).\n\n**Phase 3: Filter** → VAF range, depth, quality, PASS, variant\u002Fmutation type, consequence exclusion, population frequency, chromosome, SV size.\n\n**Phase 4: Statistics** → Type\u002Fmutation\u002Fimpact\u002Fchromosome distributions, Ti\u002FTv ratio, per-sample VAF\u002Fdepth, gene mutation counts.\n\n**Phase 5: Annotate** (optional) → MyVariant.info (ClinVar\u002FdbSNP\u002FgnomAD\u002FCADD), Ensembl VEP consequence prediction.\n\n**Phase 6: Report** → Markdown tables, direct answers, DataFrame export.\n\n**Phase 7: SV\u002FCNV Analysis** (if applicable) → gnomAD SV frequencies, ClinGen dosage sensitivity, ACMG pathogenicity classification.\n\n---\n\n## Phase Summaries\n\n### Phase 1: VCF Parsing\n\n**Use pandas for**:\n- Reading VCF as structured data\n- Quick exploratory analysis\n- When you need to manipulate columns and rows\n\n**Use python_implementation tools for**:\n- Production parsing with annotation extraction\n- Multi-sample VCF handling\n- VAF extraction from FORMAT fields\n- Large file streaming\n\n**Key functions**:\n```python\nvcf_data = parse_vcf(\"input.vcf\")           # Pure Python (always works)\nvcf_data = parse_vcf_cyvcf2(\"input.vcf\")    # Fast C-based (if installed)\ndf = variants_to_dataframe(vcf_data.variants, sample=\"TUMOR\")  # For pandas\n```\n\n### Phase 2: Variant Classification\n\n**Automatic classification from annotations**:\n- SnpEff ANN field\n- VEP CSQ field\n- GATK Funcotator FUNCOTATION field\n- Standard INFO keys: EFFECT, EFF, TYPE\n\n**Mutation types supported**: missense, nonsense, synonymous, frameshift, splice_site, splice_region, inframe_insertion, inframe_deletion, intronic, intergenic, UTR_5, UTR_3, upstream, downstream, stop_lost, start_lost\n\n**See references\u002Fmutation_classification_guide.md for full details**\n\n### Phase 3: Filtering\n\n**Common filtering patterns**:\n```python\n# Somatic-like variants\ncriteria = FilterCriteria(\n    min_vaf=0.05, max_vaf=0.95,\n    min_depth=20, pass_only=True,\n    exclude_consequences=[\"intronic\", \"intergenic\", \"upstream\", \"downstream\"]\n)\n\n# High-confidence germline\ncriteria = FilterCriteria(\n    min_vaf=0.25, min_depth=30, pass_only=True,\n    chromosomes=[\"1\", \"2\", ..., \"22\", \"X\", \"Y\"]\n)\n\n# Rare pathogenic candidates\ncriteria = FilterCriteria(\n    min_depth=20, pass_only=True,\n    mutation_types=[\"missense\", \"nonsense\", \"frameshift\"]\n)\n```\n\n**See references\u002Fvcf_filtering.md for all filter options**\n\n### Phase 4-6: Statistics, Annotation, Reporting\n\nUse python_implementation for standard stats (Ti\u002FTv, type distributions, per-sample VAF\u002Fdepth); pandas for custom aggregations. For annotation, prefer MyVariant.info (batch: ClinVar + dbSNP + gnomAD + CADD); limit to 50-100 variants per batch. Reports include type\u002Fmutation\u002Fimpact\u002Fchromosome distributions, VAF stats, clinical significance, and top mutated genes.\n\n**See references\u002Fannotation_guide.md for detailed examples**\n\n### Phase 7: Structural Variant & CNV Analysis\n\n**When VCF contains SV calls** (SVTYPE=DEL\u002FDUP\u002FINV\u002FBND):\n\n1. **Identify affected genes** (from VCF annotation or coordinate overlap)\n2. **Query ClinGen dosage sensitivity**:\n   ```python\n   clingen = ClinGen_dosage_by_gene(gene_symbol=\"BRCA1\")\n   # Returns: haploinsufficiency_score, triplosensitivity_score\n   ```\n3. **Check population frequency**:\n   ```python\n   gnomad_sv = gnomad_get_sv_by_gene(gene_symbol=\"BRCA1\")\n   # Returns: SVs with AF, AC, AN\n   ```\n4. **Classify pathogenicity**:\n   - Pathogenic: Deletion + HI score = 3, AF \u003C 0.0001\n   - Likely Pathogenic: Deletion + HI score = 2, AF \u003C 0.001\n   - VUS: HI\u002FTS score = 0-1, AF 0.001-0.01\n   - Benign: AF > 0.01\n\n**ClinGen dosage score interpretation**:\n- **3**: Sufficient evidence for dosage pathogenicity (HIGH impact)\n- **2**: Some evidence (MODERATE impact)\n- **1**: Little evidence (LOW impact)\n- **0**: No evidence (MINIMAL impact)\n- **40**: Dosage sensitivity unlikely\n\n**See references\u002Fsv_cnv_analysis.md for full SV workflow**\n\n---\n\n## Answering BixBench Questions\n\n### Pattern 1: VAF + Mutation Type Fraction\n\n**Question**: \"What fraction of variants with VAF \u003C X are annotated as Y mutations?\"\n\n```python\nresult = answer_vaf_mutation_fraction(\n    vcf_path=\"input.vcf\",\n    max_vaf=0.3,\n    mutation_type=\"missense\",\n    sample=\"TUMOR\"\n)\n# Returns: fraction, total_below_vaf, matching_mutation_type\n```\n\n### Pattern 2: Cohort Comparison\n\n**Question**: \"What is the difference in mutation frequency between cohorts?\"\n\n```python\nresult = answer_cohort_comparison(\n    vcf_paths=[\"cohort1.vcf\", \"cohort2.vcf\"],\n    mutation_type=\"missense\",\n    cohort_names=[\"Treatment\", \"Control\"]\n)\n# Returns: cohorts, frequency_difference\n```\n\n### Pattern 3: Filter and Count\n\n**Question**: \"After filtering X, how many Y remain?\"\n\n```python\nresult = answer_non_reference_after_filter(\n    vcf_path=\"input.vcf\",\n    exclude_intronic_intergenic=True\n)\n# Returns: total_input, non_reference, remaining\n```\n\n---\n\n## ToolUniverse Tools Reference\n\n### SNV\u002FIndel Annotation\n\n| Tool | When to Use | Parameters | Response |\n|------|------------|------------|----------|\n| `MyVariant_query_variants` | Batch annotation | `query` (rsID\u002FHGVS) | ClinVar, dbSNP, gnomAD, CADD |\n| `dbsnp_get_variant_by_rsid` | Population frequencies | `rsid` | Frequencies, clinical significance |\n| `gnomad_get_variant` | gnomAD metadata | `variant_id` (CHR-POS-REF-ALT) | Basic variant info |\n| `EnsemblVEP_annotate_rsid` | Consequence prediction | `variant_id` (rsID) | Transcript impact |\n\n### Structural Variant Annotation\n\n| Tool | When to Use | Parameters | Response |\n|------|------------|------------|----------|\n| `gnomad_get_sv_by_gene` | SV population frequency | `gene_symbol` | SVs with AF, AC, AN |\n| `gnomad_get_sv_by_region` | Regional SV search | `chrom`, `start`, `end` | SVs in region |\n| `ClinGen_dosage_by_gene` | Dosage sensitivity | `gene_symbol` | HI\u002FTS scores, disease |\n| `ClinGen_dosage_region_search` | Dosage-sensitive genes in region | `chromosome`, `start`, `end` | All genes with HI\u002FTS scores |\n| `ensembl_get_structural_variants` | Known SVs from DGVa\u002FdbVar | `chrom`, `start`, `end`, `species` | Clinical significance |\n\n**See references\u002Fannotation_guide.md for detailed tool usage examples**\n\n---\n\n## Common Use Patterns\n\n```python\n# Quick summary\nreport = variant_analysis_pipeline(\"input.vcf\", output_file=\"report.md\")\n\n# Filtered analysis\nreport = variant_analysis_pipeline(\"input.vcf\",\n    filters=FilterCriteria(min_vaf=0.1, min_depth=20, pass_only=True))\n\n# Annotated report (top 50 variants with ClinVar\u002FgnomAD\u002FCADD)\nreport = variant_analysis_pipeline(\"input.vcf\", annotate=True, max_annotate=50)\n```\n\n**pandas vs python_implementation**: Use python_implementation for parsing\u002Fclassification\u002Fannotation, then convert to DataFrame for custom aggregations:\n\n```python\nvcf_data = parse_vcf(\"input.vcf\")\npassing, _ = filter_variants(vcf_data.variants, criteria)\ndf = variants_to_dataframe(passing, sample=\"TUMOR\")\n```\n\n---\n\n## Limitations\n\n- **VCF annotation required for mutation classification**: If VCF has no ANN\u002FCSQ\u002FFUNCOTATION in INFO, mutation types will be \"unknown\" until ToolUniverse annotation is applied\n- **Multi-allelic variants**: Parser takes first ALT allele for type classification\n- **ToolUniverse annotation rate**: API-based, limited to ~100 variants per batch by default to respect rate limits\n- **gnomAD tool**: Returns basic metadata only (not full allele frequencies); use MyVariant.info for gnomAD AF\n- **Large VCFs**: Pure Python parser streams line-by-line; cyvcf2 is recommended for files with >100K variants\n\n---\n\n## Reference Documentation\n\n- **references\u002Fvcf_filtering.md**: Complete filter options and examples\n- **references\u002Fmutation_classification_guide.md**: Detailed mutation type classification rules\n- **references\u002Fannotation_guide.md**: ToolUniverse annotation workflows with examples\n- **references\u002Fsv_cnv_analysis.md**: Complete SV\u002FCNV interpretation workflow\n\n---\n\n## Additional Resources\n\n- Scripts: `scripts\u002Fparse_vcf.py`, `scripts\u002Ffilter_variants.py`, `scripts\u002Fannotate_variants.py`\n- Quick start recipes and MCP examples: `QUICK_START.md`\n","# Variant Analysis and Annotation\n\nProduction-ready VCF processing and variant annotation skill combining local bioinformatics computation with ToolUniverse database integration. Designed to answer bioinformatics analysis questions about VCF data, mutation classification, variant filtering, and clinical annotation.\n\n## Domain Reasoning\n\nVCF quality filtering must come before interpretation. A variant called at 2x read depth is unreliable regardless of its QUAL score, because stochastic sequencing errors at low depth can mimic true variants. The recommended minimums — depth > 10x, QUAL > 20, allele frequency consistent with expected zygosity — are not conservative; they are the floor below which calls cannot be trusted. Applying lenient filters to \"keep more variants\" sacrifices accuracy for coverage and produces false positives that propagate through all downstream analyses.\n\n## LOOK UP DON'T GUESS\n\n- Clinical significance of specific variants: query `MyVariant_query_variants` or `EnsemblVEP_annotate_rsid`; never cite ClinVar classifications from memory.\n- Population allele frequencies: retrieve from MyVariant.info or gnomAD tools; do not assume rarity.\n- ClinGen dosage sensitivity scores for genes in a CNV: call `ClinGen_dosage_by_gene`; do not estimate HI\u002FTS scores.\n- Mutation consequence predictions: run Ensembl VEP or retrieve from MyVariant.info; do not classify impact without tool output.\n\n---\n\n## CRISPR sgRNA Design Reasoning\n\n- PAM sequence (NGG for SpCas9) must lie 3' of the target on the non-target strand; the guide RNA targets the 20 nt immediately upstream of the PAM\n- For exon targeting: choose guides that cut early in the coding sequence for maximum frameshift\u002Fdisruption\n- Off-target risk increases with fewer mismatches; always check for genomic sites with 0-3 mismatches to the guide\n\n---\n\n## When to Use This Skill\n\n**Triggers**:\n- User provides a VCF file (SNV\u002Findel or SV) and asks questions about its contents\n- Questions about variant allele frequency (VAF) filtering\n- Mutation type classification queries (missense, nonsense, synonymous, etc.)\n- Structural variant interpretation requests (deletions, duplications, CNVs)\n- Variant annotation requests (ClinVar, gnomAD, CADD, dbSNP)\n- CNV pathogenicity assessment using ClinGen dosage sensitivity\n- Cohort comparison questions\n- Population frequency filtering (SNVs or SVs)\n- Intronic\u002Fintergenic variant filtering\n- Gene dosage sensitivity queries\n\n**Example Questions**:\n- \"What fraction of variants with VAF \u003C 0.3 are annotated as missense mutations?\"\n- \"After filtering intronic\u002Fintergenic variants, how many non-reference variants remain?\"\n- \"What is the clinical significance of this deletion affecting BRCA1?\"\n- \"Which dosage-sensitive genes overlap this 500kb duplication on chr17?\"\n- \"How many variants have clinical significance annotations?\"\n- \"Compare variant counts between samples\"\n\n---\n\n## Core Capabilities\n\n| Capability | Description |\n|-----------|-------------|\n| **VCF Parsing** | Pure Python + cyvcf2 parsers. VCF 4.x, gzipped, multi-sample, SNV\u002Findel\u002FSV |\n| **Mutation Classification** | Maps SO terms, SnpEff ANN, VEP CSQ, GATK Funcotator to standard types |\n| **VAF Extraction** | Handles AF, AD, AO\u002FRO, NR\u002FNV, INFO AF formats |\n| **Filtering** | VAF, depth, quality, PASS, variant type, mutation type, consequence, chromosome, SV size |\n| **Statistics** | Ti\u002FTv ratio, per-sample VAF\u002Fdepth stats, mutation type distribution, SV size distribution |\n| **Annotation** | MyVariant.info (aggregates ClinVar, dbSNP, gnomAD, CADD, SIFT, PolyPhen) |\n| **SV\u002FCNV Analysis** | gnomAD SV population frequencies, DGVa\u002FdbVar known SVs, ClinGen dosage sensitivity |\n| **Clinical Interpretation** | ACMG\u002FClinGen CNV pathogenicity classification using haploinsufficiency\u002Ftriplosensitivity scores |\n| **DataFrame** | Convert to pandas for advanced analytics |\n| **Reporting** | Markdown reports with tables and statistics, SV clinical reports |\n\n---\n\n## Workflow Overview\n\n**Phase 1: Parse VCF** → Extract CHROM\u002FPOS\u002FREF\u002FALT\u002FQUAL\u002FFILTER\u002FINFO, per-sample GT\u002FVAF\u002Fdepth, annotations (ANN\u002FCSQ\u002FFUNCOTATION). Pure Python or cyvcf2.\n\n**Phase 2: Classify** → Variant type (SNV\u002FINS\u002FDEL\u002FMNV\u002FSV), mutation type (missense\u002Fnonsense\u002Fsynonymous\u002Fframeshift\u002Fsplice\u002Fetc.), impact (HIGH\u002FMODERATE\u002FLOW\u002FMODIFIER).\n\n**Phase 3: Filter** → VAF range, depth, quality, PASS, variant\u002Fmutation type, consequence exclusion, population frequency, chromosome, SV size.\n\n**Phase 4: Statistics** → Type\u002Fmutation\u002Fimpact\u002Fchromosome distributions, Ti\u002FTv ratio, per-sample VAF\u002Fdepth, gene mutation counts.\n\n**Phase 5: Annotate** (optional) → MyVariant.info (ClinVar\u002FdbSNP\u002FgnomAD\u002FCADD), Ensembl VEP consequence prediction.\n\n**Phase 6: Report** → Markdown tables, direct answers, DataFrame export.\n\n**Phase 7: SV\u002FCNV Analysis** (if applicable) → gnomAD SV frequencies, ClinGen dosage sensitivity, ACMG pathogenicity classification.\n\n---\n\n## Phase Summaries\n\n### Phase 1: VCF Parsing\n\n**Use pandas for**:\n- Reading VCF as structured data\n- Quick exploratory analysis\n- When you need to manipulate columns and rows\n\n**Use python_implementation tools for**:\n- Production parsing with annotation extraction\n- Multi-sample VCF handling\n- VAF extraction from FORMAT fields\n- Large file streaming\n\n**Key functions**:\n```python\nvcf_data = parse_vcf(\"input.vcf\")           # Pure Python (always works)\nvcf_data = parse_vcf_cyvcf2(\"input.vcf\")    # Fast C-based (if installed)\ndf = variants_to_dataframe(vcf_data.variants, sample=\"TUMOR\")  # For pandas\n```\n\n### Phase 2: Variant Classification\n\n**Automatic classification from annotations**:\n- SnpEff ANN field\n- VEP CSQ field\n- GATK Funcotator FUNCOTATION field\n- Standard INFO keys: EFFECT, EFF, TYPE\n\n**Mutation types supported**: missense, nonsense, synonymous, frameshift, splice_site, splice_region, inframe_insertion, inframe_deletion, intronic, intergenic, UTR_5, UTR_3, upstream, downstream, stop_lost, start_lost\n\n**See references\u002Fmutation_classification_guide.md for full details**\n\n### Phase 3: Filtering\n\n**Common filtering patterns**:\n```python\n# Somatic-like variants\ncriteria = FilterCriteria(\n    min_vaf=0.05, max_vaf=0.95,\n    min_depth=20, pass_only=True,\n    exclude_consequences=[\"intronic\", \"intergenic\", \"upstream\", \"downstream\"]\n)\n\n# High-confidence germline\ncriteria = FilterCriteria(\n    min_vaf=0.25, min_depth=30, pass_only=True,\n    chromosomes=[\"1\", \"2\", ..., \"22\", \"X\", \"Y\"]\n)\n\n# Rare pathogenic candidates\ncriteria = FilterCriteria(\n    min_depth=20, pass_only=True,\n    mutation_types=[\"missense\", \"nonsense\", \"frameshift\"]\n)\n```\n\n**See references\u002Fvcf_filtering.md for all filter options**\n\n### Phase 4-6: Statistics, Annotation, Reporting\n\nUse python_implementation for standard stats (Ti\u002FTv, type distributions, per-sample VAF\u002Fdepth); pandas for custom aggregations. For annotation, prefer MyVariant.info (batch: ClinVar + dbSNP + gnomAD + CADD); limit to 50-100 variants per batch. Reports include type\u002Fmutation\u002Fimpact\u002Fchromosome distributions, VAF stats, clinical significance, and top mutated genes.\n\n**See references\u002Fannotation_guide.md for detailed examples**\n\n### Phase 7: Structural Variant & CNV Analysis\n\n**When VCF contains SV calls** (SVTYPE=DEL\u002FDUP\u002FINV\u002FBND):\n\n1. **Identify affected genes** (from VCF annotation or coordinate overlap)\n2. **Query ClinGen dosage sensitivity**:\n   ```python\n   clingen = ClinGen_dosage_by_gene(gene_symbol=\"BRCA1\")\n   # Returns: haploinsufficiency_score, triplosensitivity_score\n   ```\n3. **Check population frequency**:\n   ```python\n   gnomad_sv = gnomad_get_sv_by_gene(gene_symbol=\"BRCA1\")\n   # Returns: SVs with AF, AC, AN\n   ```\n4. **Classify pathogenicity**:\n   - Pathogenic: Deletion + HI score = 3, AF \u003C 0.0001\n   - Likely Pathogenic: Deletion + HI score = 2, AF \u003C 0.001\n   - VUS: HI\u002FTS score = 0-1, AF 0.001-0.01\n   - Benign: AF > 0.01\n\n**ClinGen dosage score interpretation**:\n- **3**: Sufficient evidence for dosage pathogenicity (HIGH impact)\n- **2**: Some evidence (MODERATE impact)\n- **1**: Little evidence (LOW impact)\n- **0**: No evidence (MINIMAL impact)\n- **40**: Dosage sensitivity unlikely\n\n**See references\u002Fsv_cnv_analysis.md for full SV workflow**\n\n---\n\n## Answering BixBench Questions\n\n### Pattern 1: VAF + Mutation Type Fraction\n\n**Question**: \"What fraction of variants with VAF \u003C X are annotated as Y mutations?\"\n\n```python\nresult = answer_vaf_mutation_fraction(\n    vcf_path=\"input.vcf\",\n    max_vaf=0.3,\n    mutation_type=\"missense\",\n    sample=\"TUMOR\"\n)\n# Returns: fraction, total_below_vaf, matching_mutation_type\n```\n\n### Pattern 2: Cohort Comparison\n\n**Question**: \"What is the difference in mutation frequency between cohorts?\"\n\n```python\nresult = answer_cohort_comparison(\n    vcf_paths=[\"cohort1.vcf\", \"cohort2.vcf\"],\n    mutation_type=\"missense\",\n    cohort_names=[\"Treatment\", \"Control\"]\n)\n# Returns: cohorts, frequency_difference\n```\n\n### Pattern 3: Filter and Count\n\n**Question**: \"After filtering X, how many Y remain?\"\n\n```python\nresult = answer_non_reference_after_filter(\n    vcf_path=\"input.vcf\",\n    exclude_intronic_intergenic=True\n)\n# Returns: total_input, non_reference, remaining\n```\n\n---\n\n## ToolUniverse Tools Reference\n\n### SNV\u002FIndel Annotation\n\n| Tool | When to Use | Parameters | Response |\n|------|------------|------------|----------|\n| `MyVariant_query_variants` | Batch annotation | `query` (rsID\u002FHGVS) | ClinVar, dbSNP, gnomAD, CADD |\n| `dbsnp_get_variant_by_rsid` | Population frequencies | `rsid` | Frequencies, clinical significance |\n| `gnomad_get_variant` | gnomAD metadata | `variant_id` (CHR-POS-REF-ALT) | Basic variant info |\n| `EnsemblVEP_annotate_rsid` | Consequence prediction | `variant_id` (rsID) | Transcript impact |\n\n### Structural Variant Annotation\n\n| Tool | When to Use | Parameters | Response |\n|------|------------|------------|----------|\n| `gnomad_get_sv_by_gene` | SV population frequency | `gene_symbol` | SVs with AF, AC, AN |\n| `gnomad_get_sv_by_region` | Regional SV search | `chrom`, `start`, `end` | SVs in region |\n| `ClinGen_dosage_by_gene` | Dosage sensitivity | `gene_symbol` | HI\u002FTS scores, disease |\n| `ClinGen_dosage_region_search` | Dosage-sensitive genes in region | `chromosome`, `start`, `end` | All genes with HI\u002FTS scores |\n| `ensembl_get_structural_variants` | Known SVs from DGVa\u002FdbVar | `chrom`, `start`, `end`, `species` | Clinical significance |\n\n**See references\u002Fannotation_guide.md for detailed tool usage examples**\n\n---\n\n## Common Use Patterns\n\n```python\n# Quick summary\nreport = variant_analysis_pipeline(\"input.vcf\", output_file=\"report.md\")\n\n# Filtered analysis\nreport = variant_analysis_pipeline(\"input.vcf\",\n    filters=FilterCriteria(min_vaf=0.1, min_depth=20, pass_only=True))\n\n# Annotated report (top 50 variants with ClinVar\u002FgnomAD\u002FCADD)\nreport = variant_analysis_pipeline(\"input.vcf\", annotate=True, max_annotate=50)\n```\n\n**pandas vs python_implementation**: Use python_implementation for parsing\u002Fclassification\u002Fannotation, then convert to DataFrame for custom aggregations:\n\n```python\nvcf_data = parse_vcf(\"input.vcf\")\npassing, _ = filter_variants(vcf_data.variants, criteria)\ndf = variants_to_dataframe(passing, sample=\"TUMOR\")\n```\n\n---\n\n## Limitations\n\n- **VCF annotation required for mutation classification**: If VCF has no ANN\u002FCSQ\u002FFUNCOTATION in INFO, mutation types will be \"unknown\" until ToolUniverse annotation is applied\n- **Multi-allelic variants**: Parser takes first ALT allele for type classification\n- **ToolUniverse annotation rate**: API-based, limited to ~100 variants per batch by default to respect rate limits\n- **gnomAD tool**: Returns basic metadata only (not full allele frequencies); use MyVariant.info for gnomAD AF\n- **Large VCFs**: Pure Python parser streams line-by-line; cyvcf2 is recommended for files with >100K variants\n\n---\n\n## Reference Documentation\n\n- **references\u002Fvcf_filtering.md**: Complete filter options and examples\n- **references\u002Fmutation_classification_guide.md**: Detailed mutation type classification rules\n- **references\u002Fannotation_guide.md**: ToolUniverse annotation workflows with examples\n- **references\u002Fsv_cnv_analysis.md**: Complete SV\u002FCNV interpretation workflow\n\n---\n\n## Additional Resources\n\n- Scripts: `scripts\u002Fparse_vcf.py`, `scripts\u002Ffilter_variants.py`, `scripts\u002Fannotate_variants.py`\n- Quick start recipes and MCP examples: `QUICK_START.md`\n",{"name":5,"description":7},"main","a298c929b6cbd662e1fce54302ce520d5ceea1a5",".agents\u002Flibrary\u002Fskills\u002Ftooluniverse-variant-analysis","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain","https:\u002F\u002Fgithub.com\u002FPrekzursil","2026-11-02 14:30:57",{"source":8,"skillName":5,"commands":40,"prompt":50,"zipAvailable":51},[41,44,47],{"tool":42,"command":43},"npx","npx skills add Prekzursil\u002Fagent-skills-toolchain --skill tooluniverse-variant-analysis",{"tool":45,"command":46},"bunx","bunx skills add Prekzursil\u002Fagent-skills-toolchain --skill tooluniverse-variant-analysis",{"tool":48,"command":49},"pnpm","pnpm dlx skills add Prekzursil\u002Fagent-skills-toolchain --skill tooluniverse-variant-analysis","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Flibrary\u002Fskills\u002Ftooluniverse-variant-analysis\u002FSKILL.md",true,[53,59,65,69,73,77,81,85,89,93,99,103],{"path":54,"resourceType":55,"resourceTypeDesc":56,"sizeBytes":57,"sourceHtmlUrl":58},"python_implementation.py","SCRIPT","脚本",65454,"https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Flibrary\u002Fskills\u002Ftooluniverse-variant-analysis\u002Fpython_implementation.py",{"path":60,"resourceType":61,"resourceTypeDesc":62,"sizeBytes":63,"sourceHtmlUrl":64},"QUICK_START.md","REFERENCE","参考文档",10101,"https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Flibrary\u002Fskills\u002Ftooluniverse-variant-analysis\u002FQUICK_START.md",{"path":66,"resourceType":61,"resourceTypeDesc":62,"sizeBytes":67,"sourceHtmlUrl":68},"references\u002Fannotation_guide.md",13217,"https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Flibrary\u002Fskills\u002Ftooluniverse-variant-analysis\u002Freferences\u002Fannotation_guide.md",{"path":70,"resourceType":61,"resourceTypeDesc":62,"sizeBytes":71,"sourceHtmlUrl":72},"references\u002Fmutation_classification_guide.md",9146,"https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Flibrary\u002Fskills\u002Ftooluniverse-variant-analysis\u002Freferences\u002Fmutation_classification_guide.md",{"path":74,"resourceType":61,"resourceTypeDesc":62,"sizeBytes":75,"sourceHtmlUrl":76},"references\u002Fsv_cnv_analysis.md",16526,"https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Flibrary\u002Fskills\u002Ftooluniverse-variant-analysis\u002Freferences\u002Fsv_cnv_analysis.md",{"path":78,"resourceType":61,"resourceTypeDesc":62,"sizeBytes":79,"sourceHtmlUrl":80},"references\u002Fvcf_filtering.md",8201,"https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Flibrary\u002Fskills\u002Ftooluniverse-variant-analysis\u002Freferences\u002Fvcf_filtering.md",{"path":82,"resourceType":55,"resourceTypeDesc":56,"sizeBytes":83,"sourceHtmlUrl":84},"scripts\u002Fannotate_variants.py",5856,"https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Flibrary\u002Fskills\u002Ftooluniverse-variant-analysis\u002Fscripts\u002Fannotate_variants.py",{"path":86,"resourceType":55,"resourceTypeDesc":56,"sizeBytes":87,"sourceHtmlUrl":88},"scripts\u002Ffilter_variants.py",8881,"https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Flibrary\u002Fskills\u002Ftooluniverse-variant-analysis\u002Fscripts\u002Ffilter_variants.py",{"path":90,"resourceType":55,"resourceTypeDesc":56,"sizeBytes":91,"sourceHtmlUrl":92},"scripts\u002Fparse_vcf.py",5702,"https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Flibrary\u002Fskills\u002Ftooluniverse-variant-analysis\u002Fscripts\u002Fparse_vcf.py",{"path":94,"resourceType":95,"resourceTypeDesc":96,"sizeBytes":97,"sourceHtmlUrl":98},"test_data\u002Fcohort2.vcf","OTHER","其他",2391,"https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Flibrary\u002Fskills\u002Ftooluniverse-variant-analysis\u002Ftest_data\u002Fcohort2.vcf",{"path":100,"resourceType":95,"resourceTypeDesc":96,"sizeBytes":101,"sourceHtmlUrl":102},"test_data\u002Fsample.vcf",6128,"https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Flibrary\u002Fskills\u002Ftooluniverse-variant-analysis\u002Ftest_data\u002Fsample.vcf",{"path":104,"resourceType":55,"resourceTypeDesc":56,"sizeBytes":105,"sourceHtmlUrl":106},"test_variant_analysis.py",35510,"https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Flibrary\u002Fskills\u002Ftooluniverse-variant-analysis\u002Ftest_variant_analysis.py",[108,129,160,187,201,217],{"skillNo":109,"name":110,"nameZh":6,"description":111,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":112,"categories":113,"keywords":114,"publishedAt":28},"01M6DF7D7478R55P46REYH4QAC","clarify-requirements","Resolve genuine pre-execution ambiguity by scanning repo context, generating 1-6 grounded options, and asking 1-3 focused questions via AskUserQuestion. NOT for mid-task use (injection risk); NOT when the goal is already unambiguous; use to disambiguate vague prompts before any file edits.","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002Ftools\u002Fworkspace-hooks\u002Fskills\u002Fclarify-requirements\u002FSKILL.md",[],[115,116,117,118,119,120,121,122,123,24,124,125,126,127,128],"clarify requirements","clarify","requirements","risk","options","task","prompts","goal","scanning","context","repo","injection","execution","file",{"skillNo":130,"name":131,"nameZh":6,"description":132,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":133,"categories":134,"keywords":140,"publishedAt":28},"01M6DF79CA4BHJS66J8EZDBPWJ","openmm-md","Run FREE, open-source, Windows-native GPU molecular dynamics via the openmm-md MCP (OpenMM + CUDA on the local NVIDIA GPU). Use for protein \u002F protein-ligand MD when you want a free stack or are on Windows without a Schrodinger license. Covers the full workflow (prepare, solvate, minimize, equilibrate, production, analyze, mmgbsa) AND an honest decision guide vs Schrodinger Desmond, including how to run Desmond legitimately on Linux or a cloud GPU with a real license. This is a deliberately LESSER free alternative and does NOT replace the schrodinger MCP tools. Do NOT use it to pick a free tool when a validated FEP+, WaterMap, IFD-MD, or Prime MM-GBSA result is actually required.","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002Ftools\u002Fmcp\u002Fopenmm-md-mcp\u002Fskill\u002Fopenmm-md\u002FSKILL.md",[135],{"categoryNo":136,"categoryName":137,"taxonomyType":138,"taxonomyTypeDesc":139},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",[141,142,143,144,145,146,147,148,149,150,151,152,153,154,24,155,156,157,158,159],"openmm md","how","decision","pick","linux","analyze","tool","stack","open","NVIDIA","gpu","license","workflow","windows","free","molecular dynamics","local","do","MCP",{"skillNo":161,"name":162,"nameZh":6,"description":163,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":164,"categories":165,"keywords":172,"publishedAt":28},"01M6DF76C1MG85VEWP8VZCBNMN","visual-review","Take screenshots of web pages and UI using Playwright for visual review and iteration","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002Ftools\u002Feditor\u002Fmetaswarm\u002Fskills\u002Fvisual-review\u002FSKILL.md",[166,169],{"categoryNo":167,"categoryName":168,"taxonomyType":138,"taxonomyTypeDesc":139},"01JBSEEDCAT000000000000003","演示文稿",{"categoryNo":170,"categoryName":171,"taxonomyType":138,"taxonomyTypeDesc":139},"01JBSEEDCAT000000000000008","测试与质量",[173,174,175,176,177,178,179,180,181,182,142,183,184,185,186],"review","visual","visual review","pages","web","false","iteration","take","check","slides","ui","screenshot","playwright","screenshots",{"skillNo":188,"name":189,"nameZh":6,"description":190,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":191,"categories":192,"keywords":193,"publishedAt":28},"01M6DF73AQVCBZKEN0YXEX9FXP","status","Diagnostic status report — shows metaswarm installation state, project setup, and potential issues","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002Ftools\u002Feditor\u002Fmetaswarm\u002Fskills\u002Fstatus\u002FSKILL.md",[],[189,194,195,196,197,198,199,200],"setup","report","project","installation","project setup","issues","state",{"skillNo":202,"name":203,"nameZh":6,"description":204,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":205,"categories":206,"keywords":208,"publishedAt":28},"01M6DF6X6C2XB0P6GX76P7TTYV","start","Use when starting work on any task, when the user mentions metaswarm, or when the user wants to begin tracked development work","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002Ftools\u002Feditor\u002Fmetaswarm\u002Fskills\u002Fstart\u002FSKILL.md",[207],{"categoryNo":136,"categoryName":137,"taxonomyType":138,"taxonomyTypeDesc":139},[203,209,120,210,211,212,24,213,214,215,216],"start issue","issue","ready","work on issue","user","work","work on","development",{"skillNo":218,"name":194,"nameZh":6,"description":219,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":220,"categories":221,"keywords":222,"publishedAt":28},"01M6DF6QMJ9AJKFKHHKRT005PQ","Interactive project setup — detects your project, configures metaswarm, writes project-local files","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002Ftools\u002Feditor\u002Fmetaswarm\u002Fskills\u002Fsetup\u002FSKILL.md",[],[194,196,157,25,198,223],"interactive"]