Unphased calls: what an array can’t say about two variants in one gene.

What phase is

Most genes come in two copies, one inherited from each parent. Phase is which variants sit together on the same copy. A genotyping array measures each position separately. It can report one copy of a variant here and one copy of another variant there, but not whether the two are on the same copy or on different ones.

When it changes the answer

For a single variant, phase makes no difference. It matters when a result depends on two variants in one gene. One copy carrying both variants, with the other copy unchanged, is a different genotype from one variant on each copy. For a recessive condition, that can be the difference between a carrier and an affected person.

Two examples

APOE types are defined by two positions read together, rs429358 and rs7412. A file that is heterozygous at both fits two combinations: ε2/ε4, which is common, and ε1/ε3, which is very rare. The usual reading is ε2/ε4, but the array’s data alone cannot rule out the other.

In HFE, one copy each of C282Y and H63D is usually read as a compound heterozygote, with the two variants on different copies. That is the common arrangement. The array does not measure it.

Pharmacogenomics

Many pharmacogenes are described by star alleles: named combinations of variants on one copy of a gene, catalogued by PharmVar. Building a pair of star alleles from unphased calls means inferring which variants travel together. When more than one pair fits the data, the pairs can predict different metabolizer types. Some star alleles are a deletion or a duplication of the whole gene, as in CYP2D6, and an array cannot see those at all.

Confirming phase

Phase can be measured by testing both parents, by long-read sequencing that spans both positions, or by an assay that reads the two variants together. Statistical phasing against reference haplotypes gives a likelihood, not a measurement.

What this means for an Atlagene report

Atlagene reads APOE and HFE with dedicated rules that take the common combination: ε2/ε4 for a file heterozygous at both APOE positions, and a compound heterozygote for one copy each of C282Y and H63D. Pharmacogenomic markers are read variant by variant. Neither phase nor copy number is measured from an array file, so where the arrangement changes the clinical reading, confirming it belongs in the laboratory’s own follow-up.