What to Do With Your Whole Genome VCF
Whole genome sequencing gives you a VCF file listing the millions of places where your DNA differs from the reference genome. Most sequencing companies report on a small fraction of it. This guide covers how to read a VCF, how to tell which file you have, what sequencing adds over a DNA chip, and what to do if you find something rare.
Updated October 2026 · Genomisaur team
What's in a VCF
A VCF (Variant Call Format) file lists variants: positions where your genome differs from the reference. A typical genome has 4 to 5 million of them, mostly single-letter changes (SNVs) plus small insertions and deletions. The file starts with header lines beginning with ##, then one line per variant:
##fileformat=VCFv4.2
##reference=GRCh38
#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT SAMPLE
chr2 135851076 rs4988235 G A 612 PASS DP=31 GT:DP 0/1:31
chr15 28120472 rs12913832 A G 1270 PASS DP=29 GT:DP 1/1:29- REF and ALTThe reference base and the alternate base found in your DNA.
- GT (genotype)0/1 means one reference and one alternate copy, 1/1 means two alternate copies, 0/0 means two reference copies. A vertical bar (0|1) means the copies are phased, so you know which came from which parent.
- FILTERPASS means the call passed the caller's quality filters. Anything else flags a lower-confidence call.
- QUAL and DPCall confidence and read depth: how many sequencing reads covered the position. Consumer WGS is usually sequenced to about 30x average depth.
One subtlety matters for analysis. A standard VCF only lists positions where you differ from the reference, so a missing position usually means you match the reference there, but it can also mean the position wasn't sequenced well. A gVCF records both, which is why gVCFs are so much larger.
To check which reference build your file uses, look for a ##reference line in the header, or check the length of chromosome 1 in the ##contig lines: 249,250,621 for GRCh37 and 248,956,422 for GRCh38. Some files name chromosomes "chr1", others just "1".
How to get your VCF ready
Most WGS providers let you download several files. Here is how to find the right one.
- 1
Find the VCF download
In your provider's dashboard, look for a downloads or raw data section. You want the file ending in .vcf.gz (or .vcf) — not .bam, .cram, or .fastq.gz.
- 2
Pick the regular VCF, not the gVCF
Some providers offer both. A gVCF (often .g.vcf.gz) includes every position in the genome and is much larger; choose the standard variant VCF instead.
- 3
Check it is a single sample
The file should contain one person. Family or trio VCFs with several people in one file are not currently supported — ask your provider for individual files.
- 4
Upload it to Genomisaur
Upload the .vcf.gz directly (no need to decompress). GRCh37 and GRCh38 are both supported. Files up to 500 MB are accepted; compressed VCFs upload much faster.
What sequencing adds over a DNA chip
- Rare variants you can trust moreSequencing reads each position directly, so rare-variant calls are far more reliable than on a chip, where most calls of very rare variants failed confirmation in one large study (Weedon et al., BMJ, 2021).
- Complete coverage for polygenic scoresNearly every variant in a score is measured directly, with no imputation needed.
- Some structural variantsDeletions and duplications can be detected with the right tools, though many consumer VCFs include only small variants.
- Medication responseSequencing data works for pharmacogenetics too. Genomisaur doesn't cover drug response; our sister site Gene2Rx accepts whole-genome VCFs and checks how your genes may affect 120 medications. Get a pharmacogenetics report on Gene2Rx
- Hard genes are still hardThe HLA genes and SMN1 (spinal muscular atrophy) need specialized analysis even with whole-genome data.
For polygenic scores specifically, the gain over a well-imputed chip is modest, because scores are built from common variants that chips and imputation already capture well. The big difference is in rare variants.
If you find something rare
With millions of variants, everyone's genome contains rare variants in disease genes. Most are harmless or of uncertain significance (often labeled VUS). ClinVar collects interpretations from clinical labs, and the American College of Medical Genetics and Genomics maintains a list of roughly 80 genes where a disease-causing variant is considered medically actionable, including BRCA1 and BRCA2, the Lynch syndrome genes, and genes for familial hypercholesterolemia.
If a tool reports a pathogenic variant in one of those genes, confirm it with a clinical test and talk to a genetic counselor. Interpretations also change as evidence accumulates, so a variant labeled uncertain today may be reclassified later.
Looking inside your VCF yourself
A compressed VCF is too large for a spreadsheet, but command-line tools handle it easily. bcftools can pull out a region or a single rsID, and Ensembl's free Variant Effect Predictor (VEP) annotates which genes your variants fall in and their predicted effect. Make sure any tool uses the same reference build as your file.
Where to take it
| Option | What you get | Price |
|---|---|---|
| Genomisaur | Polygenic risk scores for 71 conditions and traits, or a two-parent preconception report. Accepts single-sample VCFs from any provider. | $20–$29.99 plus $20 WGS processing |
| Gene2Rx | Pharmacogenetics from your VCF: how your genes may affect 120 medications. Our sister site, at gene2rx.com. | One-time; see gene2rx.com |
| Sequencing.com | Marketplace of analysis apps that accepts WGS uploads. | Free upload; apps priced individually |
| Clinical interpretation | Review of rare and medically actionable variants, usually through a doctor or genetic counselor. | Varies |
| Do it yourself | bcftools, VEP, ClinVar lookups. | Free |
Privacy before you share
A whole genome is the most identifying data you have, and it says a good deal about your relatives. Before uploading anywhere, check whether you can delete your data and whether the company shares it. In the US, GINA protects against genetic discrimination by health insurers and most employers, but not by life, disability, or long-term care insurers.
Frequently asked questions
Which sequencing providers' VCFs work?
Any provider, as long as the file is a single-sample VCF or VCF.GZ aligned to GRCh37 or GRCh38. That includes files from Nebula Genomics / DNA Complete, Dante Labs, Sequencing.com, and others.
Do I need whole genome sequencing for a good polygenic risk score?
No. Polygenic scores mostly use common variants, which consumer arrays like 23andMe and AncestryDNA already measure. WGS covers more of each score's variants directly, but the main advantage of WGS is for rare variants, which polygenic scores do not focus on.
Why is there a $20 WGS processing fee?
A whole genome VCF contains millions of variants and needs extra preparation before scoring, which uses significantly more compute than an array file.
What if I only have BAM, CRAM, or FASTQ files?
Those need variant calling first, which Genomisaur does not do. Ask your provider for a VCF export — most can generate one.
Is this medical advice?
No. Polygenic risk scores describe genetic predisposition relative to other people; they are not a diagnosis, and lifestyle and family history matter as much or more. Discuss anything concerning with a doctor or genetic counselor. Consumer arrays also miss many rare, high-impact variants, so a normal result does not rule out a hereditary condition. Most published scores, including ours, were developed in European-ancestry data, so results are less precise for people of other ancestries.
Informational, not diagnostic
Consumer genetic reports — ours, Other options's, and others — are informational. Carrier and polygenic risk results should be discussed with a qualified healthcare provider or genetic counselor before making medical or reproductive decisions.