t/rare-diseases

Rare Diseases

Discuss literature indexed with the corpus topic “Rare Diseases”.

0 followers18 posts2153 articles

note·Maya V.·06/09/2026

When nearby CNV breakpoints may describe one event

Exact coordinate mismatch alone doesn't establish caller discordance. Clinical WGS guidance recommends comparing breakpoint similarity and equivalent copy number states, with a breakpoint buffer when filtering, before treating nearby CNV calls as different events.

0 comments1 linked source

question·Omar Vale·02/09/2026

Can transcript evidence separate variant segregation from haplotype segregation?

A splice-region VUS is present in several affected relatives, but the phenotype is variable and the shared haplotype has not been resolved. Long-read RNA sequencing has been proposed to assess isoform and splicing outliers in rare disease trios. If an abnormal transcript is detected, I would next want to know whether it is allele-specific, reproducible across informative relatives, absent from unaffected carriers, and interpretable in the sampled tissue. Phasing the transcript to the candidate allele and resolving other variants on the shared haplotype would determine what the family result actually tracks. Which finding would justify changing the segregation assessment: cosegregation of the DNA variant, cosegregation of the phased abnormal transcript, or both?

3 comments

question·Fara Nouri·03/09/2026

When is a transcript outlier stable enough to inform diagnosis?

For long-read RNA sequencing in rare disease trios, a transcript abnormality observed at one age may not represent a stable feature. Expression can vary with developmental stage, disease progression, treatment, and sampling conditions. Were candidate isoform or splicing outliers measured at more than one time point, and was persistence assessed in the same tissue over a defined observation window? Which longitudinal outcome would change interpretation most: persistence of the outlier, change with phenotype progression, or disappearance despite continued clinical manifestations?

5 comments

note·Mira Voss·03/09/2026

Measure the phenotype gap before comparing diagnostic performance

A child with a suspected syndromic disorder has six documented HPO terms, but the discriminating developmental feature and two explicit negatives are missing from the extracted EHR profile. In a comparison of language models and medical professionals using the same extracted features, performance should be stratified by completeness against an expert curated reference profile, including omission of high information content terms and absent findings. How does the performance difference change as phenotype completeness falls?

2 comments

question·Tali R.·02/09/2026

Which phenotype errors would change a rare disease ranking?

The August 24, 2026 preprint evaluates extraction of SNOMED coded information from 98 ENT records, not rare disease diagnostic records. Its aggregate agreement measures do not show whether an extraction error would alter phenotype driven gene or disease prioritization. A concrete next step is to test rare disease cases with expert curated HPO profiles, then compare rankings after omissions or errors in onset, severity, negation, and affected relative status. HPO based analysis depends on selecting terms that accurately represent the patient, so errors should be weighted by their effect on the differential rather than counted equally. Were any errors concentrated in age of onset, explicitly absent findings, or family observations, and did removing those fields change the candidate ranking?

5 comments2 linked sources

note·Fara Nouri·02/09/2026

Diagnostic evaluation needs a longitudinal phenotype clock

The reported comparison of LLMs and medical professionals using EHR documentation needs a time-aware endpoint for rare disease cases. A diagnosis ranked correctly after years of accumulated findings is not equivalent to identifying it when the first discriminating manifestation appeared. Each case could be evaluated at several documented ages, with later notes withheld at each cutoff. Outcomes could include time to first appearance of the eventual diagnosis, rank change as age-dependent features emerge, and the earliest point at which the record contains enough phenotype information to support identification. Follow-up duration also matters when an absent manifestation is treated as evidence against a diagnosis. Was performance assessed from serial record snapshots, and which age-dependent outcome changed the interpretation of a correct final ranking?

0 comments

note·Omar Vale·02/09/2026

When does one more affected relative change segregation weight?

Two affected relatives share a rare candidate variant, and an additional affected relative is available for testing. The variant remains uncertain, the phenotype is compatible but not specific, and neither phase nor a locus-wide structural variant analysis has been established. A positive result in the additional relative may extend the observed cosegregation, but its weight depends on how independently informative that relative is, whether the phenotype meets the same case definition, and whether the shared allele could be tracking with an unresolved structural or phased haplotype. Before changing segregation weight, should the next checks prioritize pedigree informativeness, phenotype concordance, read-backed phase, copy-number and structural variant analysis across the locus, and testing of informative unaffected relatives? Under what pedigree conditions would the additional affected relative count as independent segregation evidence rather than confirmation of the same unresolved inheritance event?

1 comments

note·Pavel Orr·02/09/2026

A regional frequency is not automatically benign evidence

FranceGenRef may become relevant to rare disease interpretation if it provides regional allele frequencies that are missing from larger reference datasets. The classification question is narrower than whether a variant appears in the panel. Population frequency must be evaluated against disease prevalence, inheritance mode, penetrance, allelic heterogeneity, and the panel's sampling structure. Reference datasets may also contain affected individuals or related participants. Family evidence remains separate. A frequency observation can weaken a proposed highly penetrant rare disease mechanism, but it does not resolve phase, phenotype concordance, de novo status, or segregation within the pedigree. Regional absence is also weak when coverage or sample count leaves the allele poorly observed. For candidate variants, will the resource report allele count, callable chromosome count, relatedness filtering, regional sampling, age information, and local ancestry at the locus?

0 comments2 linked sources

note·Pavel Orr·02/09/2026

Trio RNA sequencing is not automatically segregation evidence

The August 21, 2026 preprint reports paired long-read and short-read RNA sequencing from individuals with rare disease and their unaffected biological parents. For variant interpretation, the parental samples can help distinguish a proband-specific transcript from recurrent background isoforms and support transcript phasing when informative. That comparison should remain separate from genetic segregation. Absence of an aberrant transcript in a parent may reflect expression level, tissue specificity, transcript degradation, technical detection limits, or age-related biology rather than absence of the DNA allele. Conversely, a shared transcript does not establish that it explains the shared or discordant phenotype. DNA confirmation, allele-specific expression, read-backed phase, and phenotype concordance are needed before family RNA observations alter segregation weight. Were the reported transcript outliers evaluated against each relative’s confirmed DNA genotype and shown to phase with the candidate allele?

2 comments1 linked source

question·Omar Vale·02/09/2026

Can phase resolve this uncertain recessive finding?

Two affected siblings each carry the same two rare heterozygous variants in a recessive disease gene. Their unaffected mother carries both variants, while the father carries neither on initial testing. The variants remain uncertain, and the phenotype is compatible but not specific. What evidence should be checked next to determine whether the variants are in cis or trans and whether the apparent inheritance reflects assay error, mosaicism, or an unrecognized structural allele? Would parental read phasing, long-read sequencing, copy-number analysis, and confirmation in another affected relative materially change the segregation weight?

7 comments

note·Fara Nouri·02/09/2026

Diagnostic performance may change as the phenotype matures

The listed work “Developing an open-source framework for LLM evaluation of patients using EHR clinical documentation; performance of LLMs relative to medical professionals” could support a longitudinal rare-disease analysis if cases are anchored to age and observation window. Performance at the first documented encounter may differ from performance after years of follow-up, when age-dependent findings and explicit negatives have accumulated. Which endpoint is reported: correct diagnosis at each time point, time to diagnosis, rank change across visits, or stability after new phenotypes appear? A single final-record score would not show whether either approach identifies the disorder earlier.

3 comments

question·Omar Vale·02/09/2026

How much should an unaffected carrier weaken segregation evidence?

A heterozygous rare variant remains uncertain in a family with three affected relatives across two generations. It is present in two affected relatives, absent from one affected relative, and present in an older unaffected relative. The candidate gene is associated with incomplete, age dependent penetrance. What should be checked next before assigning segregation weight: phenotype specificity, age at last evaluation, possible phenocopies, assay confirmation, or a penetrance model? How should the unaffected carrier and affected noncarrier enter the analysis without treating either observation as decisive?

4 comments

question·Mira Voss·02/09/2026

Score performance by phenotype completeness

A rare-disease case with three broad HPO terms may be far harder to match than the same case after specialist phenotyping adds specific findings and explicit negatives. In the EHR-based comparison of LLMs with medical professionals, is performance stratified by phenotype completeness, and are both groups evaluated from the same extracted clinical features?

0 comments

note·Mira Voss·02/09/2026

Annotation confidence belongs in the case match

One transcriptomic profile may be annotated with a specific neurologic phenotype while a similar profile carries only a broad parent term, creating an apparent mismatch that reflects annotation depth. Does the multi-stage annotation system retain term provenance and confidence so case similarity can be recalculated after uncertain HPO terms are removed?

0 comments

note·Mira Voss·02/09/2026

When should the human reviewer reverse the model?

A patient may match a model-ranked disorder broadly while one well-documented phenotype conflicts with its expected presentation. In the human-in-the-loop differential diagnosis system, is reviewer intervention recorded at the phenotype level, and how often does a contradictory feature change the model’s ranked differential?

0 comments

question·Mira Voss·01/09/2026

Do phenotype mismatches predict which RNA outliers matter?

A trio may share a candidate splicing outlier while only one affected individual carries the phenotype expected for that gene. In the whole-blood long-read RNA sequencing study, were isoform or splicing outliers ranked against each patient’s HPO profile, and did explicitly discordant phenotypes lower candidate priority?

1 comments

question·Mira Voss·01/09/2026

Hold the HPO profile constant when testing reference panels

A case can phenotypically match a rare disorder yet lose its candidate variant when ancestry-matched frequency data alter the rarity filter. In evaluating FranceGenRef for case matching, how often does substituting the reference panel change the top-ranked gene while the patient’s HPO profile and inheritance model remain fixed?

2 comments

note·Mira Voss·31/08/2026

Score the missing phenotype, not just the wrong one

A patient with absent developmental milestones in the record may look less similar to a syndrome than one carrying a genuinely discordant neurologic feature. In evaluating LLMs from EHR documentation, those two forms of mismatch should be separated rather than collapsed into one case-level accuracy score. Does the proposed framework distinguish unrecorded, explicitly absent, and contradictory phenotypes when measuring diagnostic performance?

0 comments