Skip to content

Drug Indication Priority Analysis

Generated: 2026-04-21 20:49 UTC

Summary Stats

  • Total diseases in source YAML: 3079
  • Diseases with at least one drug-indication block: 34
  • Already curated in kb/disorders/ by exact MONDO ID: 20
  • Missing from dismech by exact MONDO ID: 14
  • Missing-candidate specificity/actions: CURATE_ROOT: 11, REVIEW_SERIES_FOR_PARENT_ROOT: 2, CURATE_ROOT_WITH_SUBTYPES: 1

Top 50 Prioritized Candidates

Rank MONDO ID Disease Drug rows Medium/high evidence rows ClinGen D/S (desc) Action Priority rationale
1 MONDO:0007534 Beckwith-Wiedemann syndrome 17 44 0/0 CURATE_ROOT_WITH_SUBTYPES Broad syndromic disorder, but still worth prioritizing because the imprinting/growth-pathway biology is tractable and the treatment literature is deep.
2 MONDO:0010382 fragile X-associated tremor/ataxia syndrome 8 21 0/0 CURATE_ROOT Good mechanistic repeat-expansion disease target with enough therapeutic literature to support a focused dismech entry.
3 MONDO:0008564 DiGeorge syndrome 11 17 0/0 CURATE_ROOT Deletion syndrome with pleiotropic biology, but immune/endocrine treatment signals and a recognizable disease entity still make it a solid medium-high target.
4 MONDO:0008678 Williams syndrome 13 20 0/0 CURATE_ROOT Reasonable syndromic target with clear copy-number biology and enough cardiovascular/neurodevelopmental therapeutic literature to justify curation.
5 MONDO:0007452 maturity-onset diabetes of the young type 1 10 16 0/0 REVIEW_SERIES_FOR_PARENT_ROOT Mechanistically strong, treatment-responsive monogenic diabetes subtype, but better handled with an explicit MODY root/series strategy than as a lone subtype.
6 MONDO:0007453 maturity-onset diabetes of the young type 2 7 10 0/0 REVIEW_SERIES_FOR_PARENT_ROOT Mechanistically strong monogenic diabetes subtype with clear pharmacogenomic implications, though it should probably be curated under a MODY family plan.
7 MONDO:0011929 chromosome 1p36 deletion syndrome 27 55 0/0 CURATE_ROOT Drug-rich but heterogeneous contiguous-gene deletion syndrome; many signals are symptom-targeted rather than a clean disease-mechanism curation target.
8 MONDO:0029141 Usher syndrome, type 4 17 28 0/0 CURATE_ROOT Disease entity is valid, but the term is subtype-level and much of the current therapy discussion is inherited-retinal/Usher extrapolation rather than USH4-specific.
9 MONDO:0010775 retinitis pigmentosa-deafness syndrome 12 8 0/0 CURATE_ROOT Potentially useful sensory-disease entry, but much of the drug signal is shared retinal-disease pipeline work rather than syndrome-specific intervention evidence.
10 MONDO:0007810 autosomal dominant ichthyosis vulgaris 10 26 0/0 CURATE_ROOT Reasonable disease entity, but most current signals are topical/systemic retinoid management or broader ichthyosis extrapolation rather than a standout mechanistic target.
11 MONDO:0008698 achalasia 12 20 0/0 CURATE_ROOT Clear clinical entity, but mechanism is heterogeneous and much of the treatment signal is symptomatic or procedure-adjunctive rather than disease-mechanistic.
12 MONDO:0007523 Ehlers-Danlos syndrome, hypermobility type 9 6 0/0 CURATE_ROOT Important disorder, but weak molecular anchoring makes mechanism curation harder than for the stronger monogenic candidates above it.
13 MONDO:0007184 alopecia, androgenetic, 1 28 41 0/0 CURATE_ROOT High raw drug count, but awkward subtype granularity and mostly common alopecia treatment literature make this a weak dismech target.
14 MONDO:0012454 alcohol sensitivity, acute 9 15 0/0 CURATE_ROOT Lowest-fit candidate: the term behaves more like a susceptibility/physiologic response bucket, and its drug list mixes causes, triggers, and related-condition therapy.

Methodology Notes

  • Parsed the downloaded Monarch YAML at /tmp/prioritised-rare-disease-list.yml and used the top-level diseases list.
  • Treated a disease as having drug-indication data when it contained at least one non-empty research[].drug_label block.
  • Counted Drug rows as the number of such drug_label blocks; counted Medium/high evidence rows from nested evidence objects where confidence_drug was MEDIUM or HIGH.
  • Considered a disease already covered only when an existing kb/disorders/*.yaml file had the same disease_term.term.id MONDO CURIE.
  • Reused the repo's dismech.compare.mondo_priority heuristics to label each missing disease as CURATE_ROOT, CURATE_ROOT_WITH_SUBTYPES, or a lower-fit series/review case based on MONDO hierarchy metadata from the local sqlite DB.
  • Loaded ClinGen support from cached cache/clingen/gene_validity.csv and counted Definitive/Strong assertions on the candidate MONDO term or its descendants, which reduces false negatives for broad disease roots.
  • Final rank = quantitative score (specificity + drug-signal density + treatment rank + descendant-aware ClinGen support) plus a transparent curator-fit adjustment. Those manual adjustments mainly downgraded broad, heterogeneous, or awkwardly granular terms whose drug lists were dominated by symptomatic/general therapy rather than a clear disease-mechanism target.
  • Only 14 missing diseases met the inclusion rule, so the requested Top 50 table contains all available uncurated candidates.