domains / science / joint-variant-calling / joint-calling

joint-calling

Skill card

combined

The two skills cover the same GATK joint-calling pipeline but have distinct, non-overlapping material. The first copy is a procedural runbook with specific commands for CombineGVCFs, GenomicsDBImport, GenotypeGVCFs, and VQSR/Hard Filtering. The second copy is a conceptual guide that adds the GLnexus/DeepVariant alternative, explains the 'N+1' problem, and provides critical warnings about the difference between joint genotyping and merging VCFs (the `./.` vs `0/0` distinction). Neither copy contains the other's unique steps or constraints, so crowning one would hide a real alternative workflow (GLnexus) and important theoretical context.

Repositories

Other copies

Combined skill

---
name: bio-variant-calling-joint-calling
description: Joint genotype calling across multiple samples using GATK or GLnexus. Essential for cohort studies and VQSR.
tool_type: cli
primary_tool: GATK
---

# Joint Calling

Call variants jointly across multiple samples for improved accuracy and consistent genotyping.

## Core Principle

**Never merge single-sample VCFs.** Use gVCFs to ensure a squared-off genotype matrix where `0/0` (hom-ref) is distinct from `./.` (no data). Joint genotyping rescues low-coverage het sites by borrowing evidence across the cohort.

## Workflow Overview

1. **Per-Sample gVCF Generation**: `gatk HaplotypeCaller -ERC GVCF` (or `deepvariant --output_gvcf`).
2. **Consolidation**: 
   - Small cohorts (<100): `gatk CombineGVCFs`
   - Large cohorts (>100): `gatk GenomicsDBImport` (per chromosome for parallelism)
   - DeepVariant cohorts: `glnexus_cli --config DeepVariantWGS`
3. **Joint Genotyping**: `gatk GenotypeGVCFs` (or `glnexus_cli` for DeepVariant).
4. **Filtering**: VQSR (for >30 samples) or Hard Filtering.

## Step 1: Per-Sample gVCF Generation

```bash
gatk HaplotypeCaller \
    -R reference.fa \
    -I sample1.bam \
    -O sample1.g.vcf.gz \
    -ERC GVCF
```

## Step 2: Consolidation

### Option A: CombineGVCFs (Small Cohorts)

```bash
gatk CombineGVCFs \
    -R reference.fa \
    -V sample1.g.vcf.gz \
    -V sample2.g.vcf.gz \
    -O cohort.g.vcf.gz
```

### Option B: GenomicsDBImport (Large Cohorts)

```bash
# Create sample map
ls *.g.vcf.gz | while read f; do echo -e "$(basename $f .g.vcf.gz)\t$f"; done > sample_map.txt

# Import per chromosome
gatk GenomicsDBImport \
    --sample-name-map sample_map.txt \
    --genomicsdb-workspace-path genomicsdb_chr1 \
    -L chr1
```

### Option C: GLnexus (DeepVariant)

```bash
glnexus_cli --config DeepVariantWGS --reference reference.fa --output cohort.vcf.gz *.g.vcf.gz
```

## Step 3: GenotypeGVCFs

```bash
gatk GenotypeGVCFs \
    -R reference.fa \
    -V gendb://genomicsdb_chr1 \
    -O chr1.vcf.gz
```

## Step 4: Filtering

### VQSR (Recommended for >30 Samples)

```bash
gatk VariantRecalibrator \
    -R reference.fa \
    -V cohort.vcf.gz \
    --resource:hapmap,known=false,training=true,truth=true,prior=15.0 hapmap.vcf.gz \
    --resource:omni,known=false,training=true,truth=false,prior=12.0 omni.vcf.gz \
    --resource:1000G,known=false,training=true,truth=false,prior=10.0 1000G.vcf.gz \
    --resource:dbsnp,known=true,training=false,truth=false,prior=2.0 dbsnp.vcf.gz \
    -an QD -an MQ -an MQRankSum -an ReadPosRankSum -an FS -an SOR \
    -mode SNP \
    -O snps.recal \
    --tranches-file snps.tranches

gatk ApplyVQSR \
    -R reference.fa \
    -V cohort.vcf.gz \
    --recal-file snps.recal \
    --tranches-file snps.tranches \
    -mode SNP \
    --truth-sensitivity-filter-level 99.5 \
    -O cohort.snps.vcf.gz
```

### Hard Filtering (Small Cohorts)

```bash
gatk VariantFiltration \
    -R reference.fa \
    -V cohort.vcf.gz \
    --filter-expression "QD < 2.0" \
    --filter-name "LowQD" \
    -O cohort.filtered.vcf.gz
```