A man of twenty-six with a lifelong intellectual disability, a seizure disorder and a distinctive facial appearance is seen in an adult genetics clinic for the first time. His mother is alive and willing. His father died when he was a child, and no sample was ever stored. The laboratory's trio pathway cannot run, and the singleton pathway it falls back to returns a list of two hundred rare coding variants with no inheritance information attached to any of them.
A de novo missense variant in a gene that would fit him well is on the list. With no father, it is one of the two hundred.
What the field knows
The trio earned its place in the clinic by what it subtracts. Lee and colleagues (2014) reported a higher diagnostic rate when both parents were sequenced with the proband than when the proband was sequenced alone, and Farwell and colleagues (2015) showed the same with an analysis built around inheritance models: de novo, recessive, X-linked, each with its own filter that only parental genotypes make possible. Retterer and colleagues (2016), across a large clinical series, found that the gain from parents was greatest in the conditions where de novo variants dominate, such as developmental disorders and epilepsies.
The reason is arithmetic. A child carries a small number of new variants that neither parent has, and a very large number inherited from each parent. Parental genotypes remove the inherited ones from a dominant search in one step. Without them, the search has to fall back on indirect evidence that a variant is likely to be new and likely to matter: its absence from population databases, the constraint of its gene against variation of that class (Samocha et al., 2014; Karczewski et al., 2020), and its fit to the phenotype.
Those indirect signals are informative and they are weaker. A variant absent from a population database is consistent with being de novo and also with being a rare inherited variant from an unsampled population. A constrained gene argues that a damaging variant in it would be expected to cause disease, and says nothing about whether this patient's variant is new. The trio turns a probability into an observation. The singleton has to live with the probability.
There is a middle ground that is used less than it could be. One parent gives half the subtraction: every variant found in the mother is inherited from her and, for a dominant condition in an unaffected mother, moves down. Siblings, affected or unaffected, add segregation. A grandparent or an aunt can sometimes stand in for a missing parent for a specific variant, by showing it is present in the father's side of the family and so inherited.
Where practice breaks
Laboratory pathways are binary. There is a trio pathway and a singleton pathway, and a case with one parent falls into the singleton one, where the mother's sample, if it was even taken, is not used. The information she carries is thrown away because the pipeline has no duo mode, or has one that nobody switched on.
The missing parent is treated as a closed door. The request form records 'father deceased' and the matter ends. Nobody asks whether tissue was stored at a post-mortem, whether a newborn blood spot exists, whether the paternal grandparents are living, or whether a paternal half-sibling would consent. Each of these has resolved the inheritance of a specific variant in real cases, and each is only pursued if someone thinks of it when a candidate has emerged.
The gap itself goes unrecorded. The singleton report says that no causative variant was identified, in the same words it would use for a trio. A reader cannot tell that the analysis was weaker and why, that a plausible candidate was set aside because its inheritance could not be established, or that a single relative's sample would change the answer. The case is filed as negative, and the specific thing that would reopen it is in nobody's plan.
For adults the problem is the rule rather than the exception. Parents of a patient in their forties are often dead or estranged, and the adult clinic sees a backlog of people who were never tested as children because the tests did not exist. Pathways designed for paediatric trios serve them badly, and the diagnoses they could have, with the relatives they do have, are not pursued.
How Tulparus handles it
The case model records which samples exist and which relationships they confirm, and the analysis adapts to the family it has rather than to a template. With a mother's sample, maternal inheritance is established for every shared variant and the candidate list is filtered accordingly. With a sibling, segregation is recorded per variant. With one parent missing, every candidate carries an explicit inheritance state: maternal, not maternal and therefore paternal or de novo, or unknown.
For the man above, the missense variant would be shown as not maternal, in a constrained gene, absent from population data and fitting his phenotype. The hypothesis built on it would carry one named gap: its origin, paternal or new, is not established. Beside the gap, the workbench lists what would close it, drawn from the pedigree it has: a sample from either paternal grandparent, a paternal half-sibling, stored tissue if any exists. Each is a proposed step with what it would do to the hypothesis.
The gap stays with the case. If the mother later mentions that the father's sister is living, the step is already written and the case can be reopened on it. If a reanalysis is triggered because the gene's association is strengthened, the reanalysis starts from the hypothesis and its gap, and the person reading it sees that the inheritance question is still open.
The report says what the analysis could and could not do. It describes the samples analysed, states that paternal inheritance could not be assessed, and lists the candidate whose standing depends on it, in plain words. A clinician or counsellor reading it in five years knows exactly what to ask the family.
What you can do tomorrow
The first change is in the intake conversation, and the counsellor is usually the person who has it.
- When a parent is unavailable, ask the next questions: stored tissue, a newborn blood spot, living grandparents, half-siblings. Write the answers into the case, even when they are no.
- Use the parent you have. If your pipeline has no duo mode, ask for one; if it has one, make it the default when a trio cannot be formed.
- Report the gap. A singleton negative in a case where a trio was intended should say so, and name any candidate whose inheritance could not be established.
- Keep a list of cases waiting on a relative's sample, with the relative and the variant, so that a change in the family's circumstances can be acted on.
Sources
- Lee H, Deignan JL, Dorrani N, et al. Clinical exome sequencing for genetic identification of rare Mendelian disorders. JAMA, 2014.
- Farwell KD, Shahmirzadi L, El-Khechen D, et al. Enhanced utility of family-centered diagnostic exome sequencing with inheritance model-based analysis: results from 500 unselected families with undiagnosed genetic conditions. Genetics in Medicine, 2015.
- Retterer K, Juusola J, Cho MT, et al. Clinical application of whole-exome sequencing across clinical indications. Genetics in Medicine, 2016.
- Samocha KE, Robinson EB, Sanders SJ, et al. A framework for the interpretation of de novo mutation in human disease. Nature Genetics, 2014.
- Karczewski KJ, Francioli LC, Tiao G, et al. The mutational constraint spectrum quantified from variation in 141,456 humans. Nature, 2020.