All articles

A new gene is a patient waiting.

The literature names new genes every month, and many explain patients already sequenced. Most laboratories learn of them by chance, and the patients wait.

A paper in March described a new neurodevelopmental disorder caused by de novo variants in a gene that had, until then, no disease association. A laboratory three hundred miles from the authors held exome data on four hundred unresolved children. Three of them carried qualifying variants in the gene. Nobody looked until the following year's reanalysis round, and in the interval one of the three families had a second affected child, having been told the recurrence risk was unknown.

The answer had been on a disk in the laboratory for two years before the paper, and in the paper for eleven months before anyone read it against the cohort. Nothing about the case had been difficult. It had been unlucky in its timing, twice.

What the field knows

New gene-disease associations arrive at a pace that shows no sign of slowing. Bamshad and colleagues (2019) reviewed the rate at which Mendelian genes have been discovered since exome sequencing became routine and concluded that a large share of the remaining disorders will be solved by continued discovery rather than by any single technology. Each discovery is, by definition, an explanation for patients who were sequenced before it existed, sitting in laboratory archives with their variants already called and already set aside.

Reanalysis finds them. Wenger and colleagues (2017) showed that systematic reanalysis of negative exomes produced new diagnoses, and that the largest single contributor was gene-disease associations published since the original analysis. Liu and colleagues (2019) reported the same from a large clinical laboratory, and the Deciphering Developmental Disorders study (Wright et al., 2023) attributed a substantial share of its eventual diagnoses to reanalysis as the gene list grew. The finding is consistent across cohorts: the data were sufficient, the knowledge was not, and the knowledge arrived. What varied between services was how long it took anyone to notice, and the variation was years.

The field has built the curation infrastructure to turn a paper into an assertion a laboratory can use. The Gene Curation Coalition (DiStefano et al., 2022) harmonises gene-disease validity classifications across the groups that produce them, with evidence levels and dates, so that a laboratory can see whether an association rests on one family or many and when it was last assessed. The assertion is a thing a system can read, which is the difference between a paper and a trigger.

Where practice breaks

Laboratories learn of new genes by reading. A scientist sees a paper, or a colleague mentions it at a meeting, and remembers a patient, or does not. The mechanism is memory, and it works for the memorable cases and fails for the rest. The children with the least distinctive presentations, who most need the search, are the ones least likely to be remembered when the paper comes out.

The cohort query is a script somebody wrote. Checking four hundred exomes for variants in a newly published gene is a technical task that most laboratories can do and few do routinely, because it needs a bioinformatician's hour and the gene list nobody maintains. The query runs when someone asks, and someone asks when they remember. In a service with a long unresolved cohort and a short staff, the honest answer to how often is rarely.

Reanalysis is driven by the calendar. An annual or biennial round is better than none, and it means a gene published the month after a round waits two years. The round also reanalyses every case regardless of whether anything has changed for it, which is most of the work and most of the cost, and which is why the round happens every two years and not every month.

The family is not told to wait. A family counselled that the cause is unknown makes decisions on that basis, and nobody tells them that the search continues or what would trigger a call. When the gene is published they are not contacted, because the laboratory has not noticed, and when the laboratory notices the family may have moved, or had another child.

How Tulparus handles it

A knowledge release is diffed against every open case. When a release adds a gene-disease assertion, from a curated source with its evidence level and date, the new gene is checked against the variants held on every unresolved case, including the variants that the original analysis filtered out for sitting in a gene with no association. Each case with a qualifying variant receives a queue entry that names the gene, the assertion, its source and evidence level, the release that carried it, and the patient's variant.

The queue is triaged by evidence. An assertion resting on one family is shown as that, and the scientist decides whether to act on it or wait; an assertion with a definitive classification goes to the top. Nothing is drafted automatically. The entry is an invitation to look, with the reasons, and the look is a scientist's. A scientist who declines to act records why, and the entry closes with that reason, so that the same gene does not come back on the same evidence next month.

The case records what the look found. A reanalysis triggered by a specific gene is recorded with the gene, the release and the outcome, so that the next person knows this gene was considered in this case on this date. If the finding changes the report, the report is revised from the case record, and the decision to recontact the family is recorded with the person who made it and the date.

For the laboratory above, the three cases would have been in the queue the week the release carrying the assertion was published, each with the patient's variant and the paper's evidence beside it, and the family would have had the conversation about recurrence before the second pregnancy. The laboratory would also have been able to say, to a commissioner, how many of its open cases a year's discoveries had touched.

What you can do tomorrow

The gap between a paper and a cohort check can be closed with a list and a routine, before any software is involved. The routine is the part services lack; the list is published monthly.

  • Subscribe to a curated gene-disease validity feed and check each month's new definitive and strong assertions against your unresolved cohort. The check is one query per gene.
  • Keep the filtered-out variants. A pipeline that discards variants in unassociated genes has thrown away the material a new gene needs.
  • Tell families at the time of a negative result that the search continues, what would trigger a call, and how to keep their contact details current.
  • Record every new-gene check against the cohort, with the gene and the date, so that the next check does not repeat it.

Sources

  1. Bamshad MJ, Nickerson DA, Chong JX. Mendelian gene discovery: fast and furious with no end in sight. American Journal of Human Genetics, 2019.
  2. Wenger AM, Guturu H, Bernstein JA, Bejerano G. Systematic reanalysis of clinical exome data yields additional diagnoses: implications for providers. Genetics in Medicine, 2017.
  3. Liu P, Meng L, Normand EA, et al. Reanalysis of clinical exome sequencing data. New England Journal of Medicine, 2019.
  4. Wright CF, Campbell P, Eberhardt RY, et al. Genomic diagnosis of rare pediatric disease in the United Kingdom and Ireland. New England Journal of Medicine, 2023.
  5. DiStefano MT, Goehringer S, Babb L, et al. The Gene Curation Coalition: a global effort to harmonize gene-disease evidence resources. Genetics in Medicine, 2022.

Published in March.See the queue that fills the week it is.

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