Whole-Genome Sequencing

Whole genome sequencing reads nearly the whole of a person's DNA, about 3 billion bases, including coding regions, introns, regulatory regions and non-coding DNA. It finds single base variants, small insertions and deletions, copy number changes, structural rearrangements, repeat expansions (with specialised tools) and mitochondrial DNA variants in one test.

Compared with exome sequencing, genome sequencing has more even coverage of coding regions, a better detection of structural variants and copy number changes, and it covers deep intronic and regulatory regions. It needs no capture step, and it generates much larger data files that need heavy computing and storage.

Short read sequencing at about 30-fold average depth is the usual choice for germline testing. Long read sequencing improves detection of structural variants, repeat expansions and phasing. Tumour genomes are sequenced at greater depth, with a matched normal sample.

The turnaround is 3 to 8 weeks for routine tests and 3 to 7 days for rapid tests in critically ill infants. The diagnostic yield in rare disease is about 30 to 45%, which is 5 to 15 percentage points higher than for exome sequencing alone in some groups, mainly because of structural variants and non-coding variants.

Indications.

• Critically ill infants and children with a suspected genetic disease, where a rapid diagnosis is possible.
• Undiagnosed rare disease after a negative exome, panel or microarray.
• Suspected structural variants, repeat expansions or non-coding variants that other tests miss.
• Neurodevelopmental disorders, congenital anomalies and complex neurological disease.
• Cancer genomics, with paired tumour and normal sequencing, in selected cancers.
• Pharmacogenomic profiling in selected settings, and research programmes in newborn and population sequencing.

Not indicated when a simpler test answers the question, or when the family cannot receive counselling about the large amount of information that the test produces.

Specimen requirements.

• EDTA blood (3 to 5 mL, with high quality, high-molecular-weight DNA for long read methods), or saliva.
• Parental samples for trio analysis.
• Fresh or frozen tumour tissue and a normal sample for cancer genomes.
• Detailed phenotype, family history and previous test results.
• Consent covering secondary findings, data storage and sharing, reanalysis, and use in research.

Normal appearances.

A negative result states the depth of coverage, the types of variant analysed, and the limits. A normal genome does not exclude a genetic cause, since interpretation of non-coding variants is limited and novel disease genes are still being found.

Interpretation.

• Single nucleotide variants and small indels are filtered and classified in the same way for exome sequencing, with the addition of non-coding regions that regulate known disease genes.
• Structural variants and copy number changes (deletions, duplications, inversions, translocations) are detected and their effect on genes assessed, which can replace microarray and karyotype in some cases.
• Repeat expansions (for example, in Fragile X, Huntington disease, spinocerebellar ataxias and C9orf72) are detected with dedicated algorithms and confirmed by a targeted test.
• Mitochondrial DNA variants are covered, and heteroplasmy levels are estimated, though lower levels may be missed.
• Non-coding and deep intronic variants are classified with caution, and functional tests (RNA studies) often confirm their effect.
• Secondary findings are reported only when consent was given.
• Trio analysis finds de novo variants and phase of compound heterozygous changes.
• The result may change the diagnosis, and it supports counselling about recurrence risk, prenatal options and treatment.

Critical values and findings needing immediate action.

• A diagnosis of a treatable condition in a critically ill infant: immediate communication to the team, with changes to treatment or care.
• A pathogenic variant in a gene with preventable consequences: surveillance and risk reducing steps.
• Pathogenic results that affect a current pregnancy or relatives at imminent risk: urgent counselling.
• Findings that challenge the family structure or reveal unexpected consanguinity: handled in line with the consent given.

Limitations and artefacts.

• Interpretation of non-coding variants is difficult, since the effect of most is unknown, and many variants of uncertain significance are found.
• Short reads handle repeat regions, segmental duplications, pseudogenes and highly similar genes poorly.
• Low level mosaicism and low level mitochondrial heteroplasmy may be missed at 30-fold depth.
• Large data volumes require storage, secure transfer and bioinformatic expertise, and costs are high.
• Incidental findings are more numerous, so counselling and consent processes need to be clear.
• Data privacy and sharing: genomic data can identify individuals and relatives, and sharing for research needs clear consent.
• Detection of repeat expansions and structural variants requires confirmation by an independent method.
• Reference data for non-European ancestries remain limited, and uncertain results are more common.
• The method is not always better than exome sequencing for a given patient, and the result should add to the clinical picture.

Special groups.

• Critically ill infants: rapid genome sequencing with trio analysis can give a diagnosis in days, and it can change treatment in 15 to 30% of those tested.
• Children with undiagnosed disease after a negative exome: a genome adds new diagnoses in a minority.
• Patients with cancer: tumour and germline genomes identify drivers and hereditary risk, with specialist review.
• Newborn and population screening studies: ethical, legal and psychological issues need careful design.
• Pregnancy: prenatal genome sequencing is limited to research and specialised centres.
• Regions: genome sequencing is expensive and available at few centres, often through research collaborations and international partnerships. Capacity building for local sequencing, bioinformatics and counselling, and inclusion of local populations in reference data, are needed to make results more useful.

Follow up tests.

• Parental testing and segregation analysis.
• Targeted confirmation (Sanger sequencing, repeat primed PCR, MLPA, microarray) of key variants.
• RNA sequencing or splicing assays for non-coding and splice variants.
• Functional studies and biochemical tests.
• Long read sequencing for unresolved structural or repeat variants.
• Genetic counselling, cascade testing and specialist surveillance.
• Re-analysis of the data after a defined interval.