Version: GRCh38 1.0 Assembly: Human Dec. 2013 (GRCh38/hg38)
Description
The Genome of the Netherlands (GoNL) is a
whole-genome sequencing project covering the Dutch population. The cohort was drawn from five
Dutch biobanks and includes 250 parent-offspring families (231 trios and 19 quartets) from 11 of
the 12 Dutch provinces. Samples were not selected by phenotype or disease status. This track
shows allele counts and frequencies from the GRCh38 re-analysis of GoNL, restricted to the 498
unrelated parents (250 fathers and 248 mothers; two mothers failed QC in the original release).
The track shown here uses the GRCh38 re-analysis (version 1.0). All samples were re-aligned from
raw reads to a GRCh38 analysis set (GRCh38_no_alt_plus_hs38d1 with PhiX as decoy). The processing
pipeline is documented in the
README accompanying the data and differs from the original Nature Genetics
pipeline (reference below). Per-library reads were trimmed with cutadapt 1.13, aligned with bwa mem 0.7.15, sorted
with Picard SortSam 2.9.0, and base-quality-recalibrated with GATK BaseRecalibrator 3.7. Per-sample
files were merged and deduplicated with sambamba 0.6.6, and variants were called per sample with
GATK HaplotypeCaller 3.7. Per-family GVCFs were merged with GATK CombineGVCFs, and all families
were jointly genotyped with GATK GenotypeGVCFs 3.7. The GRCh38 callset has not been filtered with
VQSR and missing genotypes have not been imputed, so it is rougher than the original GRCh37
release.
The file
multisample.parents_only.info_only.vcf.gz was downloaded from
https://download.molgeniscloud.org/downloads/gonl_public/variants/GoNL_GRCh38_1.0/.
Of the 31,114,481 records in the source file, 30,904,161 were kept after dropping calls on the
GRCh38 decoy contigs (chrUn_JTFH01* and similar) and the EBV contig, which are not part of the
UCSC hg38 assembly. The original chromosome naming already uses the UCSC chr prefix, so
no renaming was needed. The 2,629,361 multiallelic sites were then split with
bcftools norm -m-any, with indels left-aligned against the hg38 reference, yielding
36,363,474 biallelic records (3,559,402 indels realigned). The maximum observed allele number
(AN) is 996, which matches the 498 diploid parents in the cohort. Loading documentation is in the
varFreqs makeDoc file; helper scripts for the broader varFreqs collection
are in our GitHub scripts directory.
Credits
Data was generated by the Genome of the Netherlands Consortium and distributed via the
MOLGENIS infrastructure at the
University Medical Center Groningen. Thanks to the participants who donated samples and to the
BBMRI-NL biobanks: LifeLines, Leiden Longevity Study, Netherlands Twin Registry, Rotterdam Study
and Rucphen Study.