A 30x genome reads each position about 30 times on average, so variants, including rare ones, are called directly from the data. Low-pass sequencing (around 0.5–1x) leaves much of the genome unread and uses imputation from a reference panel to infer common variants.
Low-pass works well for ancestry and polygenic scores but largely misses rare, new and structural variants. If you care about carrier status or single-gene conditions, choose 30x.
What coverage depth means
Sequencers don’t read a genome end to end. They read millions of short fragments, which software then lines up against a reference genome. Coverage, or depth, is the average number of reads that land on each position. “30x” means each base was read about 30 times on average; “0.5x” means about half the genome’s length was sequenced in total, spread thinly across it.
The word “average” is doing a lot of work. Illumina’s coverage technical note explains that reads land randomly, so in the idealized model the number of reads per position follows a Poisson distribution and many positions get fewer than the average. Run the same arithmetic at low depth and about 6 in 10 positions get no read at all at 0.5x, and roughly 4 in 10 at 1x. At 30x almost every position is read many times over, which is what lets software tell a real variant from a one-off read error and see both copies of each chromosome.
In one line: 30x sequencing reads your DNA. Low-pass sequencing samples it, then fills in the gaps statistically.
How low-pass sequencing plus imputation works
With so many positions unread, a low-pass test leans on imputation. Human DNA is inherited in long shared stretches, called haplotypes. Imputation software compares the scattered reads it has against a reference panel of thousands of fully sequenced genomes, works out which known haplotypes you most likely carry, and infers the genotypes in between.
For common variation, this works well. In a peer-reviewed study in Genome Medicine, Homburger and colleagues (2019) found imputation accuracy above 0.90 (r²) at 0.5x for all ten test genomes, and polygenic scores from low-coverage sequencing correlated closely with scores from genotyping arrays (r² of 0.93 to 0.97). A Genome Research study by Li and colleagues (2021) sequenced 1000 Genomes samples at targets of 0.5x and 1x and concluded that, at effective coverage of about 0.5x and above, low-pass sequencing plus imputation gave more accurate polygenic risk estimates than a widely used genotyping array. Its authors worked for Gencove, a company that sells low-pass analysis software, as the paper discloses.
What imputation can’t do
Imputation can only fill in what the reference panel already contains. Li and colleagues note that their imputation tool cannot call variants that are new to the panel, and the Homburger pipeline only imputed variants above 1% frequency in at least one reference population. That leaves out exactly the variants many people buy whole-genome sequencing to find:
- Rare and private variants, including most variants behind single-gene (monogenic) conditions.
- New variants that aren’t in any reference panel, including ones unique to your family.
- Small insertions and deletions and larger structural changes, which need dense reads to detect reliably.
- Consistent accuracy for everyone. Results depend on the reference panel, including how well it represents your ancestry.
An imputed genotype is a statistical best guess, even when the guess is usually right. That matters if a file is later used for health interpretation.
What 30x gets you, and its own limits
About 30x is the common benchmark for a full human genome; Illumina’s coverage guidance cites 30x to 50x for human whole-genome sequencing, depending on the application. When the 1000 Genomes Project resequenced its expanded cohort of 3,202 samples at a targeted 30x (Byrska-Bishop et al., Cell, 2022), the higher depth improved the discovery of rare single-letter variants as well as small insertions, deletions and structural variants compared with the project’s earlier low-coverage data.
30x is not perfect. Some regions of the genome are repetitive or hard to sequence with short reads, coverage is still uneven, and a consumer test is not the same as a clinical diagnostic test. A finding that could affect your health should be confirmed by an accredited clinical lab and discussed with a clinician or genetic counselor before you act on it. Our results guide explains what to do with a finding.
Side by side
| Low-pass (~0.5–1x) + imputation | 30x whole genome | |
|---|---|---|
| How genotypes are obtained | Mostly inferred from a reference panel | Mostly read directly from many overlapping reads |
| Common variants | Good, per the studies above | Good |
| Rare, new and family-specific variants | Largely missed | Detected across most of the genome |
| Insertions, deletions, structural variants | Weak | Much better (still limited in hard regions) |
| Raw files | Smaller; the VCF is largely imputed | Large BAM/CRAM, FASTQ and VCF with directly called variants |
| Suits | Ancestry, polygenic scores, common-variant traits | Carrier screening, rare-variant analysis, keeping a reusable full genome |
Which one fits what you want
Low-pass is a reasonable fit if you want ancestry estimates, common-variant trait reports or polygenic scores. Those rely on common variation, which imputation handles well. Keep in mind that polygenic scores describe statistical tendencies, not diagnoses.
30x is the one to choose if you care about rare variants: carrier status before a pregnancy, a family history of a single-gene condition, or keeping a complete dataset you can reanalyse as tools improve. Our guide to whole-genome sequencing for carrier screening covers that use in more depth. If you’re unsure, the test finder asks a few questions and points you to the right type of test.
How to tell which one you’re buying
Product pages don’t always lead with coverage, so look for these signals:
- A stated depth. “30x” or “30X average coverage” is a high-coverage genome. Figures like 0.4x, 0.5x, 1x or “low-pass” mean imputation is doing most of the work.
- The words “imputed” or “imputation”. Imputation is used with low-pass sequencing and with genotyping arrays; either way, much of the genotype data is inferred rather than read.
- “Whole genome” with no coverage number. Treat as unknown until you find it in the FAQ, the specifications or the terms.
- “Genotyping”, “array” or “chip”. That is not sequencing at all; it reads a preselected set of positions.
- Which raw files you get. A FASTQ and BAM or CRAM alongside the VCF usually indicates real depth; our what’s included page explains each file.
Questions to ask before you buy
- What is the average coverage, and is it a guaranteed minimum or a target?
- Is any part of my genotype data imputed? If so, is that marked in the files?
- Which raw files can I download (FASTQ, BAM or CRAM, VCF), and for how long?
- Is the lab accredited, and do health findings get confirmed before they’re reported?
- If my sample falls short of the stated coverage, what happens?
The comparison page sets the main test types side by side.
Commercial option · affiliate link
Want the whole genome and every raw file?
Dante Labs sequences the whole genome at 30× and lets you download the raw BAM, FASTQ and VCF files.
Check Dante’s current offer →Our code GENOME may apply; codes may not combine with sitewide sales. Why this is an affiliate link.
Whichever you choose, verify your download when it arrives and keep it somewhere safe; the files are the part of the test you keep.
Common questions
Is low-pass sequencing better than a genotyping array?
For common variants and polygenic scores, published comparisons found low-pass sequencing at about 0.5x and above performed as well as or better than a standard array after imputation. Neither is designed to find rare variants.
Can a low-pass test tell me my carrier status?
Not reliably. Many carrier variants are rare, and imputation can only infer variants that are already in its reference panel. Carrier screening calls for 30x sequencing or a clinical test, with results reviewed by a genetic counselor.
Is 30x coverage enough to find everything?
No test finds everything. 30x is the common benchmark for a full genome, but some repetitive regions are hard to read with short reads, and health findings should be confirmed by a clinical lab.
What does an imputed VCF mean for third-party tools?
Many of the genotypes are statistical estimates rather than direct reads. That's fine for ancestry-style analysis, but treat any rare-variant or health finding from imputed data with extra caution.
How can I tell what coverage my existing data has?
Check your provider's report or file documentation for a mean coverage figure. If it isn't stated, ask their support; samtools coverage also reports mean depth per chromosome from a BAM or CRAM file if you're comfortable with a terminal.
Keep going
Sources
Checked in October 2026: coverage definition and the Poisson model from Illumina’s coverage technical note, and the 30x to 50x figure from Illumina’s coverage guidance page; low-pass results from Homburger et al. 2019 (Genome Medicine) and Li et al. 2021 (Genome Research), including their stated limitations and the Gencove affiliation of the 2021 authors; 1000 Genomes high-coverage details from the Broad Institute’s summary of Byrska-Bishop et al. 2022 (Cell). The 0.5x and 1x zero-read fractions are our arithmetic from the idealized Poisson model, not measured values. We did not name the coverage of any current consumer product other than the disclosed affiliate box, because we did not verify other companies’ 2025–2026 specifications.
- Homburger et al. 2019, Genome Medicine: Low coverage whole genome sequencing and polygenic scores
- Li et al. 2021, Genome Research: Low-pass sequencing vs genotyping arrays
- Broad Institute: Researchers expand and upgrade the 1000 Genomes Project resource
- Illumina: Estimating sequencing coverage (technical note)
- Illumina: Sequencing coverage
Educational content, not medical advice. A genetic result is information, not a diagnosis; for anything medically important, talk to a qualified clinician or genetic counselor. Some outbound links are affiliate links; see the disclosure.