Can You Sequence Your DNA at Home? What Works and What Doesn't

You can collect DNA at home, and you can sequence DNA at home, but those are two different projects. Collecting a saliva or blood sample in your kitchen and shipping it to a lab gives you a 30x whole genome with a per-base error profile good enough for variant calling. Running a nanopore device on your own extracted DNA gives you a real sequencing experiment and a useful education, but it will not give you a genome you should make health inferences from. If your goal is data about yourself, send the sample out. If your goal is to learn how sequencing works, buy a MinION and accept that the output is a teaching dataset.

This page covers both paths, the file formats you should insist on, and the analysis you can do once the data lands.

The two things “sequence DNA at home” can mean

The first meaning is a mail-in kit. You spit into a tube or do a finger-stick blood collection, a lab in the United States or Europe extracts the DNA, and an Illumina NovaSeq or Complete Genomics instrument reads it. Short-read whole-genome sequencing at 30x mean coverage is the standard product. That depth is the point at which heterozygous single-nucleotide variants are called with high sensitivity across most of the mappable genome. The choice is not arbitrary. The underlying economics of high-throughput short-read platforms are what made this affordable at all ¹.

The second meaning is doing the chemistry yourself. Oxford Nanopore’s MinION Mk1B costs a few hundred dollars for the device plus roughly $500 to $900 per flow cell, and it runs off a USB-C connection to a laptop. People do use it on human DNA at home. The constraint is yield and error: a single MinION flow cell on a good run produces on the order of 10 to 30 gigabases, which is roughly 3x to 10x coverage of a 3.1 gigabase human genome. You need several flow cells to reach usable depth, plus a high-molecular-weight DNA extraction that is hard to get right from saliva.

We would not use home nanopore as a source of personal health data. We would use it to understand what a raw signal trace looks like, what basecalling does, and why read length matters for structural variants.

Why we recommend a mail-in 30x whole genome

Two reasons. The first is data quality you can reason about. A lab running a validated short-read pipeline reports mean coverage, duplicate rate, and insert size distribution. It also reports the percent of bases above 10x and 20x. You can check those numbers before you trust a single variant call. The second is that clinical-grade interpretation standards exist for whole-genome data and assume this kind of input. Published best practices for clinical whole-genome sequencing specify coverage thresholds, confirmation of candidate variants, and explicit reporting of regions that were not adequately covered ².

Coverage matters more than most people expect. Even at 30x, some regions are systematically poorly covered: GC-rich promoters, segmental duplications, the pseudoautosomal regions, and genes with close paralogs like PMS2 and SMN1. Whole-genome sequencing covers more of the genome than exome sequencing and detects copy-number and structural changes the exome misses, which is one reason cost-effectiveness analyses have increasingly favored it over targeted panels in diagnostic settings ³. It still leaves gaps, and a good report tells you where they are.

Short reads also cannot resolve repeat expansions well. If you care about long tandem repeats or complex structural rearrangements, long reads are the right tool, and you want a lab that runs PacBio HiFi or nanopore at 20x or more. That is a separate order. It is not a home project.

The files to demand, and what to do with them

Ask for three things: the raw reads, the alignment, and the variant calls. If a provider will only give you a curated web report, treat that as a consumer product. A curated report is not a dataset.

• FASTQ or unaligned BAM/CRAM. Gzipped paired-end FASTQ for a 30x human genome is roughly 40 to 60 GB per file pair. Raw reads let you realign against a different reference later, which you will want when the reference changes.
• CRAM aligned to GRCh38, with the exact reference FASTA name and MD5. CRAM is reference-compressed and runs about 15 to 25 GB at 30x, versus 60 to 90 GB for BAM. Insist on knowing whether the reference includes ALT contigs and the decoy sequence, because alignment differs.
• gVCF plus a filtered VCF. The gVCF carries reference-confidence blocks, so you can tell “homozygous reference with 28 reads” apart from “no data here.” That distinction is the single most useful thing in the file when you go looking for a specific variant.

For inspection, samtools view -c -F 0x900 sample.cram gives you primary mapped read count, and samtools coverage gives per-chromosome depth and breadth. mosdepth --by 500 --fast-mode produces genome-wide coverage in a few minutes and will show you the poorly covered regions immediately. For annotation, run Ensembl VEP with --everything --cache --assembly GRCh38 --plugin CADD --plugin SpliceAI, or bcftools csq if you want a lighter-weight consequence call. Keep the raw VCF and annotate into a copy.

If you want to reanalyze from FASTQ, the pipeline we would use is bwa-mem2 or minimap2 -ax sr for alignment, followed by samtools markdup. Then DeepVariant handles small variants and Manta or dysgu handles structural variants. On a 16-core machine with 64 GB of RAM this is an overnight job. Storage is the real cost: a personal genome with intermediates occupies several hundred gigabytes, and this scaling problem has been recognized since sequencing output began outpacing disk and network capacity ⁴.

What sequencing can and cannot tell you

Whole-genome sequencing is strong at rare, high-penetrance coding variants. It can identify the causal variant in many single-gene conditions, and it is increasingly the first-line test in clinical genetics for suspected genetic disease ⁵. It also reads pharmacogenomic loci, carrier status for recessive conditions, and mitochondrial variants.

It is weak at prediction in healthy people. A typical genome carries on the order of 4 to 5 million variants relative to the reference, and most of them have no known consequence ⁶. Reviews of whole-genome sequencing in healthy adults have consistently emphasized how often results are ambiguous and how frequently variants of uncertain significance turn up, which creates downstream testing and worry without a clear benefit ⁷. Interpretation depends on population allele frequencies and on curated evidence that changes over time, so a variant classified as uncertain today may be reclassified in either direction ⁸.

Two practical consequences. First, reclassification means you should date every interpretation you record and plan to rerun annotation every year or two. Second, any finding that looks clinically meaningful needs a clinician and an orthogonal confirmation in an accredited lab. A research-grade call on a cancer risk gene is a hypothesis. It is not a result. Genetics professionals have argued specifically against using broad genome sequencing where a targeted, validated assay answers the question, and that logic applies to your own genome as much as to screening programs ⁹.

A reasonable plan

Order a 30x short-read whole genome with raw file access. Verify coverage yourself with mosdepth before reading any report. Annotate with VEP against gnomAD allele frequencies and ClinVar, then filter aggressively. Keep only variants with population frequency below 0.1 percent, predicted loss of function, and gene-disease associations with real evidence. Everything else goes in a holding file.

Then treat the genome as the static layer it is. Sequence once and it does not change. What changes is expression, protein abundance, and metabolic state. Those are where the interesting time-series signal lives. The long-term move is to run your DNA against measurements that vary ¹⁰.

If you also want the nanopore experience, buy the device and sequence something simple first. Lambda phage or E. coli gives you a clean, small genome and fast basecalling with Dorado. It also gives you immediate feedback on whether your library prep worked. Human DNA is the hard version of the same experiment.

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. Christina A. Austin-Tse, Vaidehi Jobanputra, Denise L. Perry, et al. Best practices for the interpretation and reporting of clinical whole genome sequencing. npj Genomic Medicine, 2022. https://doi.org/10.1038/s41525-022-00295-z ↩

3. K Schwarze, J Buchanan, JC Taylor, et al. Are whole Exome and whole Genome Sequencing Approaches Cost-Effective? A Systematic Review of the Literature. Value in Health, 2018. https://doi.org/10.1016/j.jval.2018.04.677 ↩

4. M. C. Schatz, B. Langmead. The DNA data deluge. IEEE Spectrum, 2013. https://doi.org/10.1109/mspec.2013.6545119 ↩

5. Petar Brlek, Luka Bulić, Matea Bračić, et al. Implementing Whole Genome Sequencing (WGS) in Clinical Practice: Advantages, Challenges, and Future Perspectives. Cells, 2024. https://doi.org/10.3390/cells13060504 ↩

6. Michael Snyder, Jiang Du, Mark Gerstein. Personal genome sequencing: current approaches and challenges. Genes & Development, 2010. https://doi.org/10.1101/gad.1864110 ↩

7. Noralane M. Lindor, Stephen N. Thibodeau, Wylie Burke. Whole-Genome Sequencing in Healthy People. Mayo Clinic Proceedings, 2017. https://doi.org/10.1016/j.mayocp.2016.10.019 ↩

8. Hubert E Blum, Konrad Oexle. Clinical Interpretation and Implications of Whole Genome Sequencing. Journal of Next Generation Sequencing & Applications, 2014. https://doi.org/10.4172/2469-9853.1000105 ↩

9. endorsed by the European Society of Human Genetics; the P3G International Paediatric Platform; the Human Genome Organisation; and the PHG Foundation, Heidi Carmen Howard, Bartha Maria Knoppers, et al. Whole-genome sequencing in newborn screening? A statement on the continued importance of targeted approaches in newborn screening programmes. European Journal of Human Genetics, 2015. https://doi.org/10.1038/ejhg.2014.289 ↩

10. Sang Tae Park, Jayoung Kim. Trends in Next-Generation Sequencing and a New Era for Whole Genome Sequencing. International Neurourology Journal, 2016. https://doi.org/10.5213/inj.1632742.371 ↩