Where to Find Genome Data Online, and How to Get Your Own

“Genome site” usually means one of two different things, and the distinction matters before you open a browser tab. If you want the human reference genome, the consensus sequence and its annotations, the canonical sources are the UCSC Genome Browser, Ensembl, and NCBI. All are free, and all serve essentially the same underlying assembly (GRCh38, and increasingly T2T-CHM13). If you want your genome, the specific three-billion-base sequence you inherited, no website has it. That has to be generated from your DNA by a sequencing lab and delivered to you as files, typically a CRAM or BAM of aligned reads plus a VCF of variant calls, priced today in the low thousands of dollars for a clinical-grade whole genome.

This page covers both: which public resource to reach for depending on the question, and what the path looks like from a saliva or blood sample to a set of files you can query yourself.

The reference genome is a coordinate system that belongs to no person

The human reference is a mosaic assembly derived from a small number of donors, and its value is that it gives everyone a shared address space. A statement like “chr7:117,559,590 G>A on GRCh38” is only meaningful because the community agreed on a common physical map, an effort that began with explicit proposals for a shared mapping language in the late 1980s ¹ and became the coordinated sequencing program of the Human Genome Project ². Nearly every tool you will use assumes those coordinates.

Two practical consequences follow. First, assembly version is not a detail. GRCh37 (hg19) and GRCh38 (hg38) coordinates differ by millions of bases in places, and a variant list without an explicit build is close to useless. Convert with UCSC liftOver or CrossMap, and expect a small fraction of positions to fail to map. Second, the reference is a poor representation of variation, because it is one path through a population. The 1000 Genomes Project made this concrete by cataloguing variation across many populations and showing that a typical individual carries on the order of a few million sites differing from the reference ³. That is why allele-frequency databases exist alongside the assembly.

Which site for which question

The public resources overlap, so choose by task rather than habit.

For visual inspection of a locus with many annotation tracks layered on top, use the UCSC Genome Browser. Its strength is the track hub ecosystem and the Table Browser, which lets you extract an arbitrary annotation set as BED or TSV without writing an API client. UCSC is also where much ENCODE functional data is easiest to eyeball, including DNase accessibility, histone marks, and transcription factor occupancy across cell types ⁴.

For gene models, transcript structure, and programmatic annotation, use Ensembl. The gene set is versioned by release, the REST API at rest.ensembl.org is stable, and the Ensembl Variant Effect Predictor (VEP) is the annotator we reach for first. A reasonable starting invocation on a single-sample VCF:

vep -i sample.vcf.gz -o sample.vep.vcf --vcf --cache \
  --assembly GRCh38 --everything --pick_allele_gene \
  --fasta GRCh38.fa --plugin CADD,whole_genome_SNVs.tsv.gz

--cache keeps it offline and fast. The --pick_allele_gene flag collapses the transcript explosion to one consequence per allele per gene, which is what you want before any downstream filtering.

For raw sequence, assemblies, and the primary literature-linked records, use NCBI. The FTP and Datasets interfaces give you FASTA, GFF3, and assembly reports. dbSNP supplies stable rs identifiers. For population allele frequencies, use gnomAD rather than dbSNP. gnomAD reports per-population allele counts and coverage, and it is the resource that tells you a variant your report flagged is present in 4% of a population and therefore unremarkable. For clinical interpretation of specific variants, ClinVar and the ClinGen expert panels are the reference points, with the important caveat that ClinVar submissions vary in quality and you should read the review status rather than the headline classification.

Getting your own genome

The technology path is worth understanding because it determines what your files can and cannot tell you. Short-read sequencing on Illumina instruments fragments your DNA, sequences 100 to 150 bases from each end of each fragment, and aligns those reads back to the reference ⁵. Two decades of engineering drove the per-genome cost down by roughly five orders of magnitude, a trajectory documented in detail across the post-genome-project era ⁶, ⁷. What you buy today is the output of that curve.

For a whole genome, ask for 30x mean coverage or better on a 150 bp paired-end protocol. Ask for the aligned reads as well as the variant calls. The calls alone are not a complete delivery. Concretely, a complete delivery is:

• A CRAM file aligned to GRCh38, with the exact reference FASTA and its index, plus a .crai. CRAM at 30x runs roughly 15 to 25 GB, against 80 to 120 GB for the equivalent BAM.
• A per-sample gVCF or VCF from a documented pipeline (GATK HaplotypeCaller or DeepVariant), bgzip-compressed with a .tbi index.
• A quality report: mean and median coverage, percent of callable bases at 10x and 20x, duplicate rate, insert size distribution, and contamination estimate. Under 1% cross-sample contamination and above 95% of the genome callable at 10x are the numbers we look for.

Short reads miss a lot. Repeat expansions, large structural variants, and segmental duplications such as the SMN1/SMN2 region are poorly resolved by 150 bp fragments. Long-read platforms from PacBio and Oxford Nanopore address exactly this class of problem, reading tens of kilobases at a stretch ⁸. If your interest is structural variation or phasing across long haplotypes, long reads are the right instrument and short reads will quietly give you a clean-looking but incomplete answer.

Working with your files

Once you have a CRAM and a VCF, the useful work is filtering and joining against public annotation. bcftools handles nearly all of it. Count your variants by type, restrict to high-confidence calls, and intersect with a gene panel:

bcftools stats sample.vcf.gz > stats.txt
bcftools view -i 'QUAL>30 && FMT/DP>10' -f PASS sample.vcf.gz -Oz -o filt.vcf.gz
bcftools index -t filt.vcf.gz
bcftools view -R panel.bed filt.vcf.gz | bcftools query -f '%CHROM\t%POS\t%REF\t%ALT\t%INFO/CSQ\n'

Expect roughly 4 to 5 million variants against GRCh38 for a single genome, the large majority of them common and previously catalogued ³. A few dozen to a few hundred will be rare and protein-altering. Nearly all of those will have unknown consequence, and none of them should be read as a finding on your own. Variant interpretation with clinical weight requires a genetic counselor or medical geneticist. They will also consider your family history and order confirmatory orthogonal testing where warranted. Sequencing gives you a measurement, and the measurement is not a diagnosis.

The regulatory genome is where the interesting unresolved work sits. Only about 1 to 2% of the sequence encodes protein, and most trait-associated variants fall outside those regions. They sit in enhancers, promoters, and other regulatory elements catalogued by ENCODE ⁴. Predicting whether a given noncoding variant disrupts a transcription factor binding site remains hard, and motif-based predictions generate many false positives relative to measured occupancy ⁹. This is the strongest argument for pairing a genome with functional measurement, RNA expression and protein levels, rather than treating the sequence as self-interpreting ¹⁰.

Questions people also ask

How do I find my genome? You cannot look it up, because no public database holds your individual sequence. You obtain it by having a lab sequence a saliva or blood sample and return the data files, or by requesting raw data from a consumer testing company, which in the case of array-based products such as 23andMe is a genotype file covering several hundred thousand positions rather than a genome.

What exactly is a genome, and is it just DNA? Your genome is the complete DNA sequence of your cells: about 3.1 billion base pairs across 22 autosome pairs plus sex chromosomes in the nucleus, along with a small circular mitochondrial genome of 16,569 bases. It is DNA in the chemical sense, but the functional unit includes how that DNA is packaged and regulated, which is why chromatin and methylation data add information the sequence alone does not carry.

Where is the genome located in the cell? The nuclear genome sits inside the nucleus, wrapped around histones as chromatin. Mitochondria carry their own separate genome in multiple copies per organelle, inherited maternally.

How many genomes do humans have? One nuclear genome, present in two copies in nearly every cell as a maternal and a paternal haplotype, plus the mitochondrial genome at hundreds to thousands of copies per cell.

How much does it cost to get your genome tested? Research-grade 30x whole-genome sequencing with raw data delivery runs several hundred to roughly two thousand dollars, depending on turnaround and whether you receive interpretation. Clinical whole-genome sequencing ordered through a physician, with a CLIA-certified laboratory and a written report, is typically a few thousand dollars and up. Genotyping arrays cost under $200 and measure a small, preselected fraction of sites.

What is another name for a genome? In ordinary use, “the complete DNA sequence” or “the full set of genetic material.” Technical near-synonyms describe subsets: the exome is the protein-coding portion, roughly 1 to 2% of the total.

Footnotes

1. Maynard Olson, Leroy Hood, Charles Cantor, et al. A Common Language for Physical Mapping of the Human Genome. Science, 1989. https://doi.org/10.1126/science.2781285 ↩

2. P.S Harper. Mapping and sequencing the human genome. Endeavour, 1989. https://doi.org/10.1016/0160-9327(89)90035-5 ↩

3. The 1000 Genomes Project Consortium. A map of human genome variation from population-scale sequencing. Nature, 2010. https://doi.org/10.1038/nature09534 ↩ ↩²

4. The ENCODE Project Consortium. An integrated encyclopedia of DNA elements in the human genome. Nature, 2012. https://doi.org/10.1038/nature11247 ↩ ↩²

5. Ayman Grada, Kate Weinbrecht. Next-Generation Sequencing: Methodology and Application. Journal of Investigative Dermatology, 2013. https://doi.org/10.1038/jid.2013.248 ↩

6. Jeffery A. Schloss, Richard A. Gibbs, Vinod B. Makhijani, et al. Cultivating DNA Sequencing Technology After the Human Genome Project. Annual Review of Genomics and Human Genetics, 2020. https://doi.org/10.1146/annurev-genom-111919-082433 ↩

7. Elaine R. Mardis. A decade’s perspective on DNA sequencing technology. Nature, 2011. https://doi.org/10.1038/nature09796 ↩

8. Erika Check Hayden. Genome sequencing: the third generation. Nature, 2009. https://doi.org/10.1038/news.2009.86 ↩

9. Samuel Levy, Sridhar Hannenhalli. Identification of transcription factor binding sites in the human genome sequence. Mammalian Genome, 2002. https://doi.org/10.1007/s00335-002-2175-6 ↩

10. Leroy Hood, Lee Rowen. The human genome project: big science transforms biology and medicine. Genome Medicine, 2013. https://doi.org/10.1186/gm483 ↩