Rare-disease molecular diagnostic yield

Patients with suspected rare genetic conditions often reach the end of standard clinical testing without a molecular diagnosis. When initial targeted sequencing or exome tests return negative, diagnostic centres must choose whether to order additional genomic assays, including short-read whole genomes, long-read sequencing, and transcriptomic RNA analysis. A confirmed genetic cause clarifies prognosis, halts iterative diagnostic procedures, and alters clinical management. However, running stacked molecular platforms across every unresolved patient risks exhausting laboratory resources if the extra tests yield confirmed findings in only a minor fraction of unselected referrals.

Published increments for added sequencing platforms vary widely. Curated post-negative cohorts report that long-read genome sequencing adds 7%–17% diagnostic yield, whereas a broader systematic review finds 4.5%, and modelling across a full annual cohort predicts an absolute increase of 2.5%. Across 13 within-cohort studies comparing genome-wide to narrower testing, yield estimates span 18.1% to 34.2% with severe heterogeneity (I-squared 86–89%). This 3–7 fold swing reflects how patients were selected for re-testing rather than differences between sequencing instruments alone. Pre-test referral criteria themselves produce yields of 30–40%, rising to 45% when more than two clinical criteria are met. Because existing studies conflate assay capability with referral filtering, diagnostic centres cannot determine how much diagnostic gain belongs to the molecular stack versus the phenotype admission gate.

The Rare-disease molecular diagnostic yield Apodex Lab evaluates this attribution by scoring a single diagnostic centre's post-negative cohort of 300 patients who each have matched short-read genome, long-read genome, and RNA data. Three clinical geneticists apply a pre-registered five-criterion phenotype ladder to referral-time records, separating patients into tight (at least 2 criteria met) and loose (1 or fewer criteria met) groups alongside a four-rung curve across 0, 1, 2, and more than 2 criteria. The measurement delivers the confirmed diagnostic yield per platform on a single denominator, orthogonal confirmation attrition rates, and the yield contrast between tight and loose cohorts. This evidence determines whether diagnostic centres should restrict reflex multi-platform re-testing to patients meeting strict phenotype thresholds or deploy universal reflex testing across all unresolved cases.