madhavajay/1kgp-bv-all

1000 Genomes Phase 3 GRCh38 PLINK2 Dataset

Local package of the PLINK 2.0-hosted 1000 Genomes phase 3 GRCh38 files.

The dataset is stored in PLINK 2 format:

• .pgen.zst: compressed genotype data
• .pvar.zst: compressed variant metadata
• hg38_corrected.psam: sample metadata

See SOURCE.md for source and citation notes. See SCHEMA.md for machine-readable frontmatter and human-readable schema documentation. See LICENSE for data-use and license notes.

Contents

Directory layout:

README.md
SOURCE.md
LICENSE
SCHEMA.md
data/
  hg38_corrected.psam
  chr1_hg38.pgen.zst
  chr1_hg38_rs.pvar.zst
  ...
scripts/
  qc_summary.py
  plink2_qc_pca.sh
results/
  qc_sample_summary.tsv
  qc_population_counts.tsv
  qc_superpopulation_counts.tsv
  qc_variant_summary.tsv
  qc_dataset_summary.tsv

expected_files.txt is the complete file manifest.

Sample Metadata

data/hg38_corrected.psam contains one row per sample.

#IID    PAT    MAT    SEX    SuperPop    Population
HG00096 0      0      1      EUR         GBR

Columns:

• #IID: sample/participant ID
• PAT, MAT: parental sample IDs when known
• SEX: PLINK sex code, usually 1=male, 2=female, other/0=unknown
• SuperPop: super-population label
• Population: sub-population label

The .pvar.zst files contain variant metadata. The .pgen.zst files contain genotypes in the sample order defined by the .psam.

Validate Files

scripts/verify_downloads.sh

This checks:

• all manifest files are present
• there are no .aria2 partial markers
• every .zst file is a valid zstd stream
• hg38_corrected.psam is readable

Produce Basic QC Summaries

python3 scripts/qc_summary.py

This writes TSV files under results/:

• qc_sample_summary.tsv
• qc_population_counts.tsv
• qc_superpopulation_counts.tsv
• qc_file_summary.tsv
• qc_dataset_summary.tsv

The default run reports the main practical checks:

• total samples
• male/female/unknown counts
• super-population counts
• population counts
• number of .pgen.zst and .pvar.zst files
• compressed file sizes

It does not scan every variant row by default.

For per-chromosome variant row counts and estimated diploid allele slots, run:

python3 scripts/qc_summary.py --variant-counts

This adds variant_count and estimated_diploid_allele_slots to the summaries.

For exact INFO-derived allele totals (AC/AN) and SNV/non-SNV summaries, run:

python3 scripts/qc_summary.py --full-info-scan

Both variant-count modes stream every row of every .pvar.zst file and can take several minutes. Use PLINK 2 for genotype-level missingness, MAF, HWE, and LD pruning.

Deprecated older outputs may be removed and regenerated by rerunning the script. Current default outputs are:

results/qc_sample_summary.tsv
results/qc_population_counts.tsv
results/qc_superpopulation_counts.tsv
results/qc_file_summary.tsv
results/qc_dataset_summary.tsv

PLINK2 QC/PCA Example

scripts/plink2_qc_pca.sh

The script builds temporary uncompressed per-chromosome files under work_plink2_qc/, merges autosomes, applies common QC filters, performs LD pruning, and writes outputs under results/ prefixed with plink2_qc_*.

Default thresholds:

BV_GENO=0.05
BV_MIND=0.10
BV_MAF=0.01
BV_HWE_P=1e-4
BV_INDEP_WINDOW=50
BV_INDEP_STEP=5
BV_INDEP_R2=0.2

Override them on the command line:

BV_GENO=0.02 BV_MAF=0.05 scripts/plink2_qc_pca.sh

Requirements:

• zstd
• plink2

The PLINK2 workflow is intentionally an example pipeline. Review the filter choices for your cohort and analysis goal before publishing results.