All articles

Consanguinity changes the search.

Related parents shift the odds towards recessive disease and leave a signature in the genomeSequencing of all of a person's DNA, coding and non-coding regions alike.. Both should shape the analysis. Neither settles it.

A couple who are first cousins have a daughter with a severe early-onset epilepsy. The exome is analysed under the laboratory's standard pipeline, which ranks all variant classes together. A de novo missense variant in a well-known epilepsy gene rises to the top, is classified as likely pathogenic, and is reported. The family is counselled on a recurrence risk appropriate to a new mutation.

A second child, born two years later, has the same condition and does not carry the variant. The reanalysis finds a homozygous loss-of-function variant in a recessive gene, inside a long run of homozygosity, in both children. It had been in the first exome, outside the top twenty.

What the field knows

Parental relatedness changes the prior probability of recessive disease and leaves a physical trace. A child of first cousins is homozygous by descent across a measurable fraction of the genome, in long contiguous runs, and a rare recessive variant inherited from a shared ancestor will sit inside one of them. Homozygosity mapping, which uses those runs to narrow the search for a recessive gene, predates exome sequencing (Lander and Botstein, 1987) and remains one of the strongest tools for the recessive case, since it reduces the candidate space from the whole exome to a few regions.

Large sequencing studies in populations with high rates of consanguinity have shown how much the search changes. Alkuraya (2013) described a diagnostic approach in which runs of homozygosity are combined with exome data as a matter of routine, with a high share of diagnoses in recessive genes, many of them novel at the time. Monies and colleagues (2017), reporting on thousands of families, found the same pattern, and showed that the recessive share rose with the degree of relatedness. The DDD study, in a largely outbred population, found that recessive diagnoses were a modest share overall and concentrated in the families of related parents (Martin et al., 2018).

Relatedness is also often unknown to the laboratory. Families do not always volunteer it, forms do not always ask, and endogamous communities can carry distant relatedness that nobody in the family would describe as consanguinity. The genome shows it anyway: runs of homozygosity can be measured from the proband's own data (Pedersen and Quinlan, 2017; Narasimhan et al., 2016), and the estimate is a better guide to the analysis than the pedigree form.

What relatedness does not do is settle the case. A child of related parents can still have a de novo dominant condition, and the de novo variant in the opening case was a reasonable candidate on its own. Relatedness changes the weights and the order of the search; it should never close the alternatives. The counselling consequences of getting this wrong are large, since the recurrence risk for a recessive condition in a consanguineous couple is one in four for every pregnancy, and the risk quoted for a de novo variant is far lower.

Where practice breaks

The pipeline is the same for every family. A ranking tuned for the typical outbred referral, where de novo dominant variants dominate the yield, applies the same weights to a child of first cousins, where the prior favours recessive disease. The de novo candidate rises as it would in any case; the homozygous candidate, in a less famous gene, sits lower than its prior deserves.

Runs of homozygosity are computed and not used. Many pipelines emit them as a quality metric, or for the relatedness check, and stop there. The information that a rare homozygous variant sits inside a long run, which is exactly the signal homozygosity mapping was built on, is not carried into the ranking. The scientist may know the family is consanguineous and still see a list in which that fact has changed nothing.

The pedigree form is believed over the data. If the form says the parents are unrelated, the analysis proceeds as if they were, even when the proband's genome shows runs of homozygosity that no outbred child would carry. The discrepancy is noticed, if at all, after the case has been reported. Families from endogamous communities are the most affected, because the relatedness is real, distant, and invisible to a question about cousins.

A plausible dominant candidate ends the search. Once a likely pathogenic de novo variant is found, the recessive hypothesis is not pursued, because the case appears solved. In a consanguineous family the recessive hypothesis deserved to be kept open until the dominant one had been confirmed by something more than its rank, and the second child paid for the difference.

How Tulparus handles it

Relatedness is estimated from the data on every case and recorded as a finding beside the pedigree. Where the estimate and the form disagree, the case shows both and asks the clinician to confirm with the family; the analysis proceeds on the data. Runs of homozygosity are computed for the proband and kept on the case as regions, with their lengths.

The recessive candidate channel uses them. A homozygous rare variant inside a long run of homozygosity is raised, with the run named as the reason, and the channel's output is shown beside the other channels so that the dominant and recessive hypotheses sit on the same page with their own evidence. The prior from relatedness is applied to the comparison explicitly, and the page says what it was and where it came from.

Neither hypothesis is closed by the other. A likely pathogenic de novo variant becomes a hypothesis with its standing; the homozygous candidate remains a hypothesis with its own; the comparison shows what would separate them, which in the opening case was segregation in a future sibling or functional evidence for the dominant variant. The report, rendered from the case, would have carried the recessive candidate as an open alternative with its reason, and the counselling conversation would have been different.

At reanalysis, a new sibling's data are added to the family and both hypotheses are re-evaluated against them. The run-of-homozygosity regions, the relatedness estimate and the earlier reasoning are all still there, because they were findings on the case and not lines in a quality log.

What you can do tomorrow

The first two steps use data the pipeline already produces.

  • Compute runs of homozygosity for every proband and look at them before ranking. A child with long runs is a recessive case first, whatever the form says.
  • For a consanguineous family, review the homozygous rare variants inside the runs as a separate list, before the general ranking.
  • When a dominant candidate is found in a consanguineous family, write the recessive alternative into the report as open, with what would settle it.
  • Ask about relatedness in a way that includes community and distant kinship, and record the answer with the data-derived estimate beside it.

Sources

  1. Lander ES, Botstein D. Homozygosity mapping: a way to map human recessive traits with the DNA of inbred children. Science, 1987.
  2. Alkuraya FS. The application of next-generation sequencing in the autozygosity mapping of human recessive diseases. Human Genetics, 2013.
  3. Monies D, Abouelhoda M, AlSayed M, et al. The landscape of genetic diseases in Saudi Arabia based on the first 1000 diagnostic panels and exomes. Human Genetics, 2017.
  4. Martin HC, Jones WD, McIntyre R, et al. Quantifying the contribution of recessive coding variation to developmental disorders. Science, 2018.
  5. Pedersen BS, Quinlan AR. Who's who? Detecting and resolving sample anomalies in human DNA sequencing studies with Peddy. American Journal of Human Genetics, 2017.
  6. Narasimhan VM, Hunt KA, Mason D, et al. Health and population effects of rare gene knockouts in adult humans with related parents. Science, 2016.

Related parents, two hypotheses.See both kept open on one page.

A demo takes 20 minutes, screen shared, on synthetic cases. Bring the question from one of your own.