Eight hundred and twenty-two consecutive paediatric patients had clinical exome sequencing at a single laboratory. A definitive molecular diagnosis was established in 181 of them, 22 per cent, while 40 per cent were left holding a variant of uncertain significance. A commercial AI platform sat inside the analysis pipeline, scoring variants against each proband's Human Phenotype Ontology terms during the initial case review. Of the 152 pathogenic or likely pathogenic variants in the fully resolved cases, it flagged 98.7 per cent, and ranked 75 per cent inside its top ten.
What those numbers describe is prioritisation, not diagnosis. Flagging a variant is not classifying it, and the 40 per cent uncertain rate is untouched by how well the ranking performed. The prioritisation also ran on phenotype terms, so its usefulness is bounded by how completely the phenotype was captured before sequencing began. The distance between narrowing the search and making the call runs through the rest of this edition.
Exome as a first-tier investigation across 202 individuals from 196 Indian families, with consanguinity in 19 per cent. A molecular diagnosis in 25 per cent of families, and only 60 per cent of those were in autoinflammatory genes. The rest were other inborn errors of immunity or monogenic mimics.
Variants in 515 genes have been published as causal in cerebral palsy. Treating cerebral palsy as a phenotypic feature that certain disorders raise the risk of, rather than a diagnosis to be displaced, the null hypothesis of no association could be rejected for only 89 of them.
Fifty-three studies and more than 13,000 patients. Median survival was 29 years with home mechanical ventilation against 19 years without, improving progressively over time in both groups.
Longitudinal swallowing assessments in 130 participants enrolled between 2006 and 2024. Silent aspiration arrives with no cough, no throat clearing and no wet voice to warn anyone, and aspiration pneumonia is a leading cause of death in NPC1.
GRADE-based guidance on who should be assessed for a hereditary polyposis syndrome, when germline testing is appropriate, and how endoscopic and surgical management follows from the result. Presymptomatic diagnosis in families with a known variant is where cancer incidence actually moves.
Sixty-six studies, 4,452 cases from 23 countries, pooled detection rate 19 per cent. Of the pathogenic variants found, 76 per cent sat in genes on the ACMG secondary findings list, so most of what postmortem testing recovers is already actionable for surviving relatives.
A four-tier gated framework for transcript choice. Annotate as standard, escalate only where transcript context could materially change the molecular consequence, and reserve RNA or functional evidence for what stays unresolved.
Most predictors read conservation as functional constraint, and so absorb mutation rate variation without meaning to. Variants at low-mutation-rate sites come out looking more damaging than they are, and variants at highly mutable sites more tolerated.
Maps what has been built for paediatric rare disease diagnosis, on which data modalities, and how far validation has gone. Evidence maturity is the question worth reading it for.
