Version: v3.1.2 Assembly: Human Dec. 2013 (GRCh38/hg38)
Description
A reprocessed callset by the gnomAD project combining the 1000 Genomes and Human Genome Diversity Project
(HGDP) data, with 4,094 whole genomes from 80 populations. The dataset includes per-population
allele frequencies for all 80 populations as well as broad continental groupings from gnomAD
(African, Admixed American, East Asian, European, Middle Eastern, South Asian, and others).
This track shows allele frequencies only. The full phased genotype data with haplotype
clustering display is available in the
gnomAD HGDP+1000G track under Phased Variants.
The track here does not include the full variant frequencies for all subpopulations, instead,
it aggregates frequencies to the main groups, AFR, AMI, AMR, ASJ, EAS, FIN, MID, NFE, OTH, SAS.
To access the full frequency information, use the track under "Phased Variants".
Data Access
The data can be explored interactively with the
Table Browser or the
Data Integrator.
For programmatic access, our REST API can be used; the
track name is hgdp1kFreq.
For bulk download, the VCF file can be obtained from
our download server.
The original VCFs with full genotypes can also be downloaded from
gnomAD Downloads.
Methods
The gnomAD project reprocessed 4,094 whole genomes from the 1000 Genomes Project and the Human
Genome Diversity Project (HGDP) through a unified pipeline. Sequencing was performed on Illumina
platforms at a mean coverage of 32–34x. Reads were aligned to GRCh38 (hs38DH reference with
decoy and HLA sequences) using BWA-MEM 0.7.15. Variant calling followed GATK best practices:
per-sample calls with GATK 3.5 HaplotypeCaller, then joint genotyping with GATK4 through
the Hail VCF combiner, which scales the merge step. Allele-specific variant quality score recalibration
(AS-VQSR) was applied for both SNPs and indels. Sample QC included contamination estimates
(verifyBamID), sex concordance, relatedness filters (PC-Relate), and population assignment
with PCA against gnomAD reference panels. Per-population allele frequencies were computed for
80 fine-grained populations and for broad continental groups.
The makeDoc file documents how all source files of the varFreqs track were converted.
For some tracks, python scripts were also needed and are available from GitHub.
Credits
Thanks to the gnomAD team at the Broad Institute for harmonizing and making this dataset
publicly available, and to all participants of the 1000 Genomes Project and the Human Genome
Diversity Project.