Polygenic scores in clinic: prediction requires calibration and context
A high percentile is not an absolute risk, and strong association does not establish that using a score improves care.
The reading room
Explore medical genetics
Find a starting point, a clinical perspective, or a closer look at the research.
Research & discovery
Explore methods, emerging technologies and the evidence behind discovery.
12 articles
Long-read sequencing: match the technology to the unresolved mechanism
Longer molecules can clarify structural variation and difficult regions, but analytical validation and study design still determine diagnostic value.
Single-cell genomics: more cells do not replace more participants
Cell atlases reveal biological heterogeneity, but meaningful disease comparisons require sound sampling, batch handling and independent replication.
Functional variant assays: abnormal biology is not automatically pathogenicity
Assay validation, disease mechanism and clinical calibration determine whether a laboratory observation can strengthen variant classification.
CRISPR screens: from a selected hit to a defensible mechanism
A screen identifies candidates under particular experimental conditions; orthogonal evidence is needed before treating a hit as a disease mechanism or therapeutic target.
DNA methylation episignatures: biomarkers with a defined scope
A reproducible methylation pattern can support a diagnosis, but classifier coverage and validation determine how much it can resolve.
From GWAS signal to mechanism: why the nearest gene is not the answer
Association identifies a genomic neighborhood; causal variants, target genes and biological effects require additional evidence.
Population reference datasets: rarity, constraint and ancestry in context
Reference databases are essential for variant interpretation, but absence, group labels and gene-level constraint each require careful qualification.
Gene–disease validity: establish the association before classifying a candidate
A plausible variant cannot compensate for an uncertain relationship between the gene and the proposed disorder.
Rare-disease data sharing: make data useful without treating access as unlimited
FAIR principles and responsible governance can support discovery while preserving consent, privacy and accountability.
Gene therapy evidence: distinguish early response from durable benefit
A one-time administration still requires long-term evidence about effectiveness, late adverse events and who remains represented in follow-up.
Rare-disease trial endpoints: measure a change that matters
Small populations make endpoint selection more demanding, not less; connect measurement quality, natural history and patient relevance.