Reading a genome report

A fixed order helps when reading. First: what was technically measured, for example genotype, HLA type, star allele or variant. Second: how reliable the measurement is, for example through coverage, quality, no-call or tool limits. Third: which database or guideline interprets the finding. Fourth: what follows in practice. Many misunderstandings arise when layer three or four is read without checking layer one and two.

A report can show that a variant is present or that a risk model is higher. That is not the same as a disease. Penetrance, family history, symptoms, lab values, age, ancestry and other genes change the meaning. A report should therefore prepare questions, not make decisions on its own.

For single markers it matters which allele a study treats as the effect allele. The effect allele can increase risk, decrease risk or only be associated with a measurement. Providers and databases also sometimes use different DNA strands. An apparently different letter can therefore be the same variant written in the opposite direction.

A missing entry does not automatically mean that a variant is absent. No-call, low coverage, difficult gene regions, pseudogenes or repeats can prevent a tool from deciding reliably. Good reports therefore distinguish not found, not measurable and not evaluated.

ClinVar and ACMG terms describe the current evidence base. Pathogenic or likely pathogenic means that there is solid evidence for disease relevance in the right context. VUS means that the variant is known but its meaning is unclear. Benign means that current evidence treats the variant as not disease-causing. These classes can change when new data appear.

A useful companion translates the report into questions: Is the measurement quality at this locus sufficient? Does the finding fit my family history or symptoms? Does it need confirmation with a clinical test? Is there a guideline or only an association study? Does the finding change prevention, medication choice or further diagnostics today?