Whole-genome sequencing vs SNP arrays.
Every consumer DNA test uses one of these two technologies. They're not the same: not in depth, not in coverage, not in what you can do with the data.
| Feature | Mosaic WGS (30×) | SNP arrays (consumer chips) |
|---|---|---|
| What the technology reads | All 3 billion base pairs across all chromosomes. | ~600,000 pre-selected SNP positions (~0.02% of the genome). |
| How positions are selected | No selection — every base is sequenced. | Manufacturer pre-decides which SNPs the chip captures at chip-design time. |
| Read depth | 30× — each position is read ~30 times for variant-call reliability. | Single-pass hybridization signal; not depth-based. |
| Variant types detected | SNVs, indels, structural variants, copy-number variants, novel/rare variants. | Pre-selected common SNPs only. Rare/novel variants are invisible. |
| Performance on rare variants | Detected directly from the read data. | Mostly missed unless the chip happened to include the position. |
| Imputation | Not required — direct reads. | Required to estimate non-chip positions from chip data and reference panels. |
| Reanalysis when science updates | Re-interpret the existing raw genome at no extra sequencing cost. | Limited to positions the original chip captured. |
| Throughput / cost trade-off | Higher cost per sample. Higher information density. | Lower cost per sample. Capped information ceiling. |
| Clinical regulatory standard | Run for clinical interpretation in a CLIA-, COLA-, CAP-, and AABB-accredited lab. | Consumer-grade genotyping; clinical use requires separate validation. |
What the technology reads
Mosaic WGS (30×)
All 3 billion base pairs across all chromosomes.
SNP arrays (consumer chips)
~600,000 pre-selected SNP positions (~0.02% of the genome).
How positions are selected
Mosaic WGS (30×)
No selection — every base is sequenced.
SNP arrays (consumer chips)
Manufacturer pre-decides which SNPs the chip captures at chip-design time.
Read depth
Mosaic WGS (30×)
30× — each position is read ~30 times for variant-call reliability.
SNP arrays (consumer chips)
Single-pass hybridization signal; not depth-based.
Variant types detected
Mosaic WGS (30×)
SNVs, indels, structural variants, copy-number variants, novel/rare variants.
SNP arrays (consumer chips)
Pre-selected common SNPs only. Rare/novel variants are invisible.
Performance on rare variants
Mosaic WGS (30×)
Detected directly from the read data.
SNP arrays (consumer chips)
Mostly missed unless the chip happened to include the position.
Imputation
Mosaic WGS (30×)
Not required — direct reads.
SNP arrays (consumer chips)
Required to estimate non-chip positions from chip data and reference panels.
Reanalysis when science updates
Mosaic WGS (30×)
Re-interpret the existing raw genome at no extra sequencing cost.
SNP arrays (consumer chips)
Limited to positions the original chip captured.
Throughput / cost trade-off
Mosaic WGS (30×)
Higher cost per sample. Higher information density.
SNP arrays (consumer chips)
Lower cost per sample. Capped information ceiling.
Clinical regulatory standard
Mosaic WGS (30×)
Run for clinical interpretation in a CLIA-, COLA-, CAP-, and AABB-accredited lab.
SNP arrays (consumer chips)
Consumer-grade genotyping; clinical use requires separate validation.
A SNP chip is a thumbnail. WGS is the full image.
SNP arrays are a brilliant cost-engineering trick: pre-choose ~600K of the most commonly-varied positions in the human genome and read just those, very cheaply. The chip can’t tell you what’s at the other 99.98% of your DNA.
For ancestry estimates and well-studied common-variant traits, the chip is enough. For clinical interpretation that handles rare variants, structural variants, and new science as it lands, you need the read data behind the calls — which is what 30× whole-genome sequencing gives you.
What this difference actually changes
Depth + breadth = variant calls you can act on. SNP-chip data isn’t built for that.
A SNP chip reads roughly 600,000 hand-picked positions out of 3 billion base pairs, at low depth, with significant imputation. That’s enough for ancestry estimates. It is not enough for clinical-grade variant calls in the genes that shape your detox capacity, your methylation cycle, your thyroid conversion, your iron handling, or your caffeine clearance. Whole-genome sequencing at 30× depth changes the question from “what does the chip happen to cover” to “what does your biology actually say.”
