How to Get Your DNA Sequenced, and What You Receive

“Get DNA sequence” means two different things, and the answer depends on which you want. If you want the sequence of a known gene, say the reference coding sequence of BRCA1, you download it from NCBI or UCSC in about thirty seconds and pay nothing. If you want your own genome, you order whole-genome sequencing from a clinical or direct-to-consumer laboratory. You send saliva or blood and receive between 60 and 120 GB of data a few weeks later. Consumer whole-genome sequencing currently runs roughly $300 to $1,500 depending on coverage depth and whether the laboratory is CLIA-certified. Clinical diagnostic sequencing ordered through a physician typically bills between $1,000 and $5,000 and may be covered by insurance when there is a documented indication. This page covers both, with emphasis on the second, because the file formats and quality parameters are where people get burned.

Deciding what to order: array, exome, or whole genome

The three common products differ by how much of your genome they observe, and the differences are larger than the price differences suggest.

A genotyping array (what 23andMe and AncestryDNA sell) does not sequence anything. The array queries roughly 600,000 to 1 million predetermined positions by hybridization and reports which allele you carry at each. That is about 0.03% of your 3.1 billion base pairs. Arrays are cheap and accurate at common variants, and they are the right tool for ancestry and for polygenic scores built on common SNPs. They cannot detect a variant that was not designed onto the chip, which means most rare and private variants are invisible. Array output is a flat text file of rsID, chromosome, position, and genotype, usually a few tens of megabytes.

Whole-exome sequencing captures and sequences the protein-coding regions, roughly 1 to 2% of the genome, at high depth (typically 100x mean). It catches the majority of variants currently known to cause monogenic disease, since those concentrate in coding sequence, but capture efficiency varies across targets and coverage drops in GC-rich first exons.

Whole-genome sequencing reads everything, including introns, regulatory regions, and mitochondrial DNA, and gives more uniform coverage than capture-based exomes because no hybridization step is involved. It also detects structural variants and copy-number changes that exomes miss. This is what we would order. Short-read WGS at 30x mean depth is the standard, and the reasoning behind that number is straightforward: to call a heterozygous site confidently you need several reads supporting each allele, and Poisson sampling at 30x leaves only a small fraction of the genome below the 10x threshold where genotype calls become unreliable ¹. At 15x, a meaningful share of heterozygous sites are called as homozygous reference, which is the failure mode that matters because it silently produces false negatives rather than obvious errors.

If you can get long reads (PacBio HiFi or Oxford Nanopore), take them. Long reads resolve segmental duplications, repeat expansions, and phasing (which variants sit on the same chromosome copy) that short reads cannot. The cost is higher and short-read variant calling in unique sequence is still marginally more accurate, so for most people 30x short-read WGS is the sensible default.

What files to insist on before you pay

This is the single most important thing to settle before ordering. Many consumer sequencing companies return a PDF report and a limited variant list, and treat the underlying data as an upsell or withhold it entirely. Ask for the following, in writing:

• FASTQ files, gzipped and paired-end, with two files per lane named _R1 and _R2. These are the raw reads with per-base quality scores. They let you realign to a different reference or re-call variants with a newer pipeline years later. Expect roughly 100 GB for 30x.
• BAM or CRAM, with the index (.bai or .crai). This is reads aligned to a reference. CRAM is reference-compressed and roughly half the size of BAM. Confirm which reference build: GRCh38 is current, and files aligned to hg19/GRCh37 will need liftover, which introduces errors at a small number of positions.
• gVCF as well as VCF. A plain VCF lists variant sites only. A gVCF also records the confidence that non-variant positions are homozygous reference, so you can distinguish “this position matches the reference” from “this position had no coverage.” Without it you cannot tell a true negative from missing data, so a plain VCF is not enough.

Verify what you received before the support window closes. Run samtools flagstat sample.cram and check that the mapping rate is above 98% and duplicates are below about 15%. Run samtools coverage or mosdepth --by 1000 to confirm mean depth matches what you paid for and to see the fraction of the genome above 10x, which should be well over 90%. Then bcftools stats sample.vcf.gz: a 30x human WGS should yield roughly 4 to 5 million variants relative to GRCh38, with a transition/transversion ratio near 2.0 genome-wide and a heterozygous/homozygous ratio near 1.5 to 2.0. Numbers far outside those ranges mean the pipeline, the sample, or the contamination level is off.

Cost, and why the “$1,000 genome” number misleads

The $1,000 genome was a target set in the mid-2000s when sequencing a human cost in the tens of millions, and it referred to reagent and instrument cost for producing raw sequence ². Consumables for 30x short-read WGS are now well below that figure, which is why consumer offerings can price in the hundreds. The full cost of a sequencing result in a clinical setting is considerably higher, because it includes the following.

• sample handling
• library preparation labor
• instrument amortization
• storage
• computational analysis
• variant interpretation by a certified laboratory

A detailed Dutch cost analysis found that these downstream components dominate, and that the headline sequencing price captures only part of what a laboratory spends per case ³. The economic evaluations of genomic sequencing similarly note that most published cost figures are not comparable because they measure different parts of the workflow ⁴.

The practical implication: a $399 consumer genome and a $3,000 clinical genome are not the same product. The cheaper one gives you data. The more expensive one gives you data plus a report from a laboratory director who has curated the variants against clinical evidence and will stand behind the call. If you intend to act on a finding medically, you need the second, or you need the first confirmed by the second. Variants found in a research-grade or consumer pipeline should be re-tested in a CLIA-certified clinical laboratory before any medical decision, and interpreted with a genetic counselor or physician.

Insurance and how to order clinically

In the United States, insurers generally cover sequencing when there is a documented clinical indication. Examples include a child with an undiagnosed condition suggestive of a genetic cause, a family history meeting criteria for hereditary cancer syndrome panels, or a pharmacogenomic test tied to a specific prescribing decision. Coverage for sequencing a healthy adult out of curiosity is rare. The path is a referral to a medical geneticist or genetic counselor, who documents the indication and orders from a clinical laboratory. Prior authorization is common, and the ordering clinician’s documentation determines whether it is approved.

Retrieving a reference gene sequence

If what you want is the sequence of a gene rather than your own genome, the tooling is mature and free. From NCBI, search the gene symbol in the Gene database to get the RefSeq accession. For a BRCA1 transcript that accession is NM_007294. Then fetch it:

efetch -db nuccore -id NM_007294 -format fasta > brca1.fa

efetch comes from NCBI’s Entrez Direct command-line tools. For genomic coordinates rather than a transcript, UCSC’s Table Browser or the REST endpoint at https://api.genome.ucsc.edu/getData/sequence?genome=hg38;chrom=chr17;start=43044294;end=43125482 returns the sequence for an interval. Ensembl BioMart is the better choice when you want many genes at once with a consistent annotation set.

To extract your own sequence at those coordinates from your CRAM, use samtools consensus -r chr17:43044294-43125482 or bcftools consensus applied to the reference FASTA with your VCF, which substitutes your called variants into the reference backbone.

A short note on where sequencing came from

The techniques underlying all of this were developed by Sanger and by Maxam and Gilbert in the mid-1970s, and the first-generation chain-termination method remained dominant through the Human Genome Project ⁵. The shift to massively parallel short-read platforms, and later to nanoball and single-molecule approaches, drove per-base cost down by many orders of magnitude over about fifteen years ⁶⁷. That trajectory is what makes an individual genome an ordinary purchase rather than a national research program ⁸. It is also why the interpretation layer, not the sequencing, is now the binding constraint.

Questions people also ask

How much does it cost to get DNA sequenced? Genotyping arrays run $50 to $200, consumer whole-genome sequencing at 30x runs roughly $300 to $1,500, and clinical diagnostic whole-genome sequencing ordered through a physician typically bills $1,000 to $5,000. The spread reflects depth, certification, and how much interpretation is included rather than differences in the chemistry ³.

Will insurance cover genome sequencing? Usually only with a documented clinical indication such as an undiagnosed suspected genetic condition or a family history meeting established criteria. Elective sequencing of a healthy adult is generally paid out of pocket. Ask your insurer for their specific coverage policy and CPT codes before ordering.

How do I retrieve a gene sequence from GenBank? Find the RefSeq accession through NCBI Gene, then use efetch -db nuccore -id <accession> -format fasta, or download from the web interface. Use the NM_ accession for the messenger RNA transcript and the NG_ accession for the genomic reference sequence.

What is a gene sequence? It is the ordered string of the four DNA bases across a region of the genome that encodes a functional product. The letters A, C, G, and T stand for those bases. In practice you will work with one of three forms. A coding sequence is the exons joined together, beginning with ATG. A transcript sequence adds untranslated regions, and a genomic sequence includes introns and flanking regulatory DNA.

How do I find my own sequence at a specific position? Align your reads to GRCh38, then query the resulting CRAM directly with samtools tview or extract an interval consensus with samtools consensus. Always check depth at that position first, because a region with two reads will give you a confident-looking answer that means nothing.

Footnotes

1. Rachel L Goldfeder, Dennis P Wall, Muin J Khoury, et al. Human Genome Sequencing at the Population Scale: A Primer on High-Throughput DNA Sequencing and Analysis. American Journal of Epidemiology, 2017. https://doi.org/10.1093/aje/kww224 ↩

2. Simon T Bennett, Colin Barnes, Anthony Cox, et al. Toward the $1000 Human Genome. Pharmacogenomics, 2005. https://doi.org/10.1517/14622416.6.4.373 ↩

3. Kirsten J M van Nimwegen, Ronald A van Soest, Joris A Veltman, et al. Is the $1000 Genome as Near as We Think? A Cost Analysis of Next-Generation Sequencing. Clinical Chemistry, 2016. https://doi.org/10.1373/clinchem.2016.258632 ↩ ↩²

4. Kurt Christensen, Dmitry Dukhovny, Uwe Siebert, et al. Assessing the Costs and Cost-Effectiveness of Genomic Sequencing. Journal of Personalized Medicine, 2015. https://doi.org/10.3390/jpm5040470 ↩

5. C. A. Hutchison. DNA sequencing: bench to bedside and beyond. Nucleic Acids Research, 2007. https://doi.org/10.1093/nar/gkm688 ↩

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. Gregory J Porreca. Genome sequencing on nanoballs. Nature Biotechnology, 2010. https://doi.org/10.1038/nbt0110-43 ↩

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