Young Girl ReadingJean-Honoré Fragonard, c. 1769, National Gallery of Art, Washington

A girl so taken by her book that the room has gone. The articles here are written to be read like that: one problem, what is known about it, and what to do.

Unresolved cases,examined in writing.

Long-form articles for clinical geneticists, genetic counsellors and variant scientists. Each one starts from a case that stayed open, reads what the field knows about the problem, and shows how Tulparus handles it.

15 November 20265 min read

The cheapest test is a better phenotype.

A girl of nine is referred for exome sequencing with three terms on the request form: global developmental delay, hypotonia, dysmorphic features. Her clinic letters, read together, also mention nystagmus noted at eight months and never mentioned again, a heart murmur called innocent on one echocardiogram, feeding difficulty through infancy, and a growth chart that crossed two centile lines downward between three and six. None of that reached the laboratory.

Yield follows the quality of the phenotype more than the depth of sequencing. Recording what was examined, what was absent and what was never looked at.For clinicians · For genetic counsellors
RSS feed

Phenotype

What reaches the laboratory from the clinic, and what is lost on the way.

15 November 20265 min readThe negative finding nobody wrote down.An examined and normal finding argues against a whole class of candidates. Most request forms have nowhere to put it, so the laboratory never sees it.For clinicians · For genetic counsellors15 November 20265 min readThe clinical question belongs in the case record.A test is ordered to answer a question. When it is not written down, the laboratory answers a different one, and nobody can tell when the case is done.For clinicians · For genetic counsellors15 November 20265 min readWhy possible hypotonia should stay possible.Clinicians hedge for good reasons. A record that strips the hedge and stores a firm finding feeds the ranking something the examiner never claimed.For clinicians · For genetic counsellors15 November 20265 min readAge of onset narrows the search.When a feature first appeared separates conditions that share the same terms. Forms keep the feature and lose the when, and the ranking cannot use it.For clinicians · For laboratories15 November 20265 min readWhen two diagnoses explain one patient.A patient can have two genetic conditions at once, and the blend fits neither. An analysis that stops at the first good answer misses the second.For clinicians · For laboratories15 November 20265 min readWhat the referral letter leaves out.A request form carries a line or two. The clinic letters carry the patient. A search run on the form alone uses a fraction of what the hospital knows.For laboratories · For clinicians

Variants

The classes and positions a first pass sets aside.

Ranking

How candidates are found, scored and explained.

15 November 20265 min readOne ranking channel is one point of failure.Every prioritisation method has blind spots set by its inputs. A case ranked by one method inherits them all. Several, fused and shown apart, do not.For laboratories · For clinicians15 November 20265 min readA novel gene needs a different standard.When the best candidate has no disease association, the classification rules do not apply. A case can still pursue it, if the standard is written first.For laboratories · For clinicians15 November 20265 min readExpression tells you where to look.A gene never switched on in the affected tissue is a weak candidate however rare its variant. Expression belongs in the ranking, with its limits stated.For laboratories · For clinicians15 November 20265 min readModel organism knockouts in a human differential.A mouse with the patient's features is evidence for a gene no human study has reached. It belongs in the ranking, scored as what it is.For laboratories · For clinicians15 November 20265 min readSimilar patients, without sharing patients.The strongest evidence for a rare diagnosis is often another patient like this one. Finding them can be done without moving anyone's record.For laboratories · For clinicians15 November 20265 min readSimilarity earns a look. Then the work begins.A high similarity score says a gene's phenotype resembles the patient's. It says nothing about the variant, and a number shaped like a verdict is believed.For laboratories · For clinicians15 November 20265 min readThe paralogue and the gene next door.When the phenotype points at a gene and the gene is clean, the answer is sometimes in its family. A known-gene filter removes the relatives first.For laboratories · For clinicians15 November 20265 min readThe reranker should explain itself.A learned model moved a gene from fourteenth to second. If the only answer to why is that it weighed many features, it has no place in a clinical record.For laboratories · For clinicians

Evidence

Which version of the knowledge was used, and why it matters.

15 November 20265 min readWhich ClinVar did you look at?ClinVar changes with every release. A classification quoted without its date and submitter cannot be checked, and the case cannot be reanalysed.For laboratories15 November 20265 min readProvenance is a clinical safety feature.Every fact in an interpretation came from somewhere, at some version. Keeping the where and the when is what lets a conclusion be checked when it matters.For laboratories · For clinicians15 November 20265 min readImmutable evidence, versioned interpretation.The facts in a case do not change. What they mean does. A system that edits records in place loses the first and cannot track the second.For laboratories15 November 20265 min readThe literature search nobody can repeat.Every interpretation leans on a literature search done once, in a browser, and never recorded. The next scientist starts again and may find less.For laboratories15 November 20265 min readGene-disease validity comes first.Before a variant can be pathogenic for a condition, the gene has to be established as causing it. That step is skipped more often than any report admits.For laboratories · For clinicians15 November 20265 min readWhen two laboratories disagree about a variant.The same variant is pathogenic in one submission and uncertain in another. The disagreement is information, and one label in its place throws it away.For laboratories · For genetic counsellors15 November 20265 min readThe ontology changed under the case.A phenotype is coded in terms, and the terms belong to a release. When the release moves, a case coded three years ago means something slightly different.For laboratories · For clinicians

Operations

The service around the investigation: turnaround, isolation, governance.

15 November 20265 min readTurnaround time is set by the slowest handover.Sequencing takes days. Cases take months. The difference is queues between people, and the rapid programmes show what happens when the queues are removed.For laboratories15 November 20265 min readTenant isolation is a clinical requirement.A genome identifies a person for life. A platform holding many services' cases has to keep each one's data apart at every layer, and be able to prove it.For laboratories15 November 20265 min readThe case lives in email and spreadsheets.The laboratory system holds the sample and the report. Everything between them, candidates, queries and decisions, lives in files owned by individuals.For laboratories15 November 20265 min readThe cost of a case that stays open.An unresolved case is charged to no budget line, so it looks free. The family, the clinic and the laboratory each pay for it every year.For laboratories · For clinicians15 November 20265 min readDecision support, and where the decision stays.A tool that proposes and a person who decides works only if the record shows which was which. Most systems blur the line, and always in one direction.For laboratories · For clinicians15 November 20265 min readPatient data trains nobody's model by default.The terms say de-identified data may improve the service. For genomic data that clause is larger than it looks, and the default should run the other way.For laboratories · For genetic counsellors15 November 20265 min readBenchmarks you can audit.Two vendors claim a yield and neither names the set it was measured on. A figure without its evaluation set, sample size and interval is only a sentence.For laboratories · For clinicians15 November 20265 min readFrom pilot to practice.Most clinical software pilots succeed and most are never adopted. The gap is rarely the tool. It is the governance and the exit nobody wrote down first.For laboratories · For clinicians

Start with one casestill waiting for answers.