Repository navigation
Review and deepen Ectodermal Dysplasia and Immunodeficiency 2 (MONDO:0012806) - #14737
Conversation
|
Claude encountered an error after 1s —— View job I'll analyze this and get back to you. |
|
Claude finished @cmungall's task in 2m 26s —— View job Review of PR #14737: changes requested
Summary checklist
🟡 Important (blocking)
🔵 Suggestions: add a co-trimoxazole
|
![ai4c-reviewer[bot]](https://avatars.githubusercontent.com/in/4388239?s=60&v=4){kind=small}
There was a problem hiding this comment.
Review: Ectodermal Dysplasia and Immunodeficiency 2 (PR #14737)
Changes requested. Three items need fixing before merge:
- The new Sparse Hair evidence quotes a sentence that does not mention hair.
- The trigger node is bound to the generic
GO:0005515protein binding. The PR body saysGO:0019904. - Several infection and GI phenotypes are missing, although cached references already contain quotable sentences for them.
Everything else is sound. All deterministic checks pass.
- Schema validation passes (
just validaterun locally) - All PMIDs verified as real (all cached; titles checked)
- Evidence snippets are exact quotes (43/43 verified)
- Ontology terms correct and specific: the molecular function binding is too generic (item 2)
- evidence_source classifications correct
- No pathophysiology bundling
Scope: 2 files (the entry plus its history record). No out-of-scope changes and no reverted cache files.
Checks run locally: just validate, just validate-terms, just count-verified-snippets (43/43) and just check-causal-targets all pass. just check-genereviews reports NO_CHAPTER, so no GeneReviews baseline is required.
🟡 IMPORTANT
1. The Sparse Hair evidence does not mention hair (phenotypes → Sparse Hair). The added snippet is "Thirteen patients had developmental features of EDA, the severity and nature of which differed between cases." It supports ectodermal dysplasia in general, not sparse hair, so the explanation ("including sparse scalp hair") claims more than the quote says. A cached reference names the feature directly. PMID:28629746 contains:
"Heterozygous mutations in the IκB family member IκBα cause autosomal dominant anhidrotic ectodermal dysplasia with immune deficiency (AD EDA-ID), characterized by sparse hair, conical teeth, reduced number of sweat glands, and susceptibility to severe infections."
This sentence comes from the paper's introduction, so mark it quote_role: BACKGROUND. The same sentence also fixes the older Hypohidrosis item, which reuses the same general EDA snippet.
2. The trigger node's molecular function is too generic, and it does not match the PR description. The node Constitutive NF-kB Inhibition by Gain-of-Function IkBa binds GO:0005515 protein binding. GO itself discourages that term because it says almost nothing. The PR body and the history record both say GO:0019904 protein domain specific binding with INCREASED, but the file contains neither. The specific term for what mutant IκBα does here is GO:0051059 NF-kappaB binding (checked via runoak -i ols:go). Use that term. modifier: INCREASED fits a hypermorphic inhibitor. Then update the history details to match what was actually written.
3. Infection and GI phenotypes are missing (cross-check against the Falcon deep-research report). The report lists chronic mucocutaneous candidiasis, Pneumocystis infection and chronic diarrhoea as recurrent features. None of these is in the entry, and each has a quotable sentence in a reference the entry already cites:
- Mucocutaneous candidiasis. PMID:15337789: "Except for intermittently recurrent mucosal candidiasis that responded to fluconazole, no subsequent serious infections were encountered during follow-up."
- Pneumocystis pneumonia. PMID:15337789: "In bronchoalveolar lavage fluid, Pneumocystis carinii was identified."
- Chronic diarrhoea. PMID:35005117: "She presented recurrent fever, infectious pneumonia and chronic diarrhea with EDA-ID." This sentence is already cited for Recurrent Infections.
- Absent γ/δ T cells. The PMID:28597146 sentence already quoted under Lymphocytosis states "a lack of circulating γ/δ T cells", but no phenotype records it.
Look up each HPO term before binding it, and wire each new phenotype from Impaired T Cell Development and Function or Defective Innate Immune Signaling.
🔵 SUGGESTIONS
- Antimicrobial Prophylaxis: add
therapeutic_agentfor co-trimoxazole (CHEBI:3770"co-trimoxazole", verified). The evidence quoted for this treatment names that drug, and CLAUDE.md recommends an agent whenever the treatment term isNCIT:C15986. - Gene-disease validity: the CGGV assertion is cited but not recorded. Add
gene_disease_validity(DEFINITIVE/CLINGEN/external_id) on the NFKBIA record;just list-gene-validitywill list it asbackfill. - Non-infectious inflammation: the deep-research report describes steroid-responsive systemic and CNS inflammation in some Ser32 variants (Moriya 2018). That paper is not cached, so this is not blocking. It would make a good follow-up node.
- Diagnosis evidence: the snippet shows the mutation causes the disease, not that genetic testing is how it is diagnosed. That is acceptable given there is no GeneReviews chapter.
What looks good
- The new causal edges are correct, and every phenotype is now connected to the pathograph.
- Renaming the phenotype to Decreased Circulating B Cell Count removes its name clash with the B Cell Deficiency pathophysiology node.
- The stromal and lymph-node mechanism is correctly graded
MODEL_ORGANISM. - Prevalence (
CASES_IN_LITERATURE+ULTRA_RARE), progression and the NEMO differential are well supported. - Every new evidence item carries a title. I spot-checked 4 titles against the cache and all match.
…plasia-and-immunodeficiency-2
|
Updated
|
|
Claude finished @cmungall's task in 3m 48s —— View job Re-review of PR #14737: changes requested
Summary checklist
Fixed since the first review: the Sparse Hair and Hypohidrosis evidence now quotes a sentence naming those features, marked 🟡 Still to fix (small)
🔵 Suggestions: make the category spelling consistent ( |
There was a problem hiding this comment.
Re-review: Ectodermal Dysplasia and Immunodeficiency 2 (PR #14737)
Changes requested, but only small ones remain. Most of the first review is fixed: the Sparse Hair and Hypohidrosis evidence now quotes the PMID:28629746 sentence that names the features, the four requested phenotypes are added and wired into the pathograph, and the ClinGen tier and co-trimoxazole agent are recorded. Three problems are left: the four new phenotypes claim FREQUENT on case-report evidence, the history record describes phenotypes the file does not contain, and the trigger node uses a parent GO term instead of the specific one.
- Schema validation passes (
just validate) - All PMIDs verified as real (all cached; 56 titles checked, 0 issues)
- Evidence snippets are exact quotes (48/48 verified)
- Ontology terms correct and specific (item 3)
- evidence_source classifications correct
- No pathophysiology bundling
Checks run locally: just validate, just validate-terms, just count-verified-snippets (48/48) and just check-causal-targets all pass. check-gene-activity-grounding reports findings only in other entries, not this one. Scope is still 2 files.
🟡 IMPORTANT
1. The evidence does not support frequency: FREQUENT on the four new phenotypes. Chronic Mucocutaneous Candidiasis, Pneumocystis jirovecii Pneumonia and Chronic Diarrhea are each backed by one patient in one case report (PMID:15337789 or PMID:35005117). The gamma-delta T cell snippet says the feature was found "in some, but not all patients". In an ultra-rare disease with about 14 published patients, a single case does not establish a frequency band. Remove frequency from all four. If you can quote a cohort count, you could use OCCASIONAL for gamma-delta.
2. The history record and the PR comment describe content that is not in the file. The history details (and the 06:34 PR comment) say the entry adds Severe Combined Immunodeficiency (HP:0004430), Mycobacterial Infections (HP:0020102) and Impaired Polysaccharide Antibody Response (HP:0500271). What the file actually adds is Chronic Mucocutaneous Candidiasis (HP:0002728), Pneumocystis jirovecii Pneumonia (HP:0020102), Chronic Diarrhea (HP:0002028) and Decreased Gamma-Delta T Cell Proportion (HP:0500271). Two of the CURIEs in the record are attached to the wrong labels. History records are permanent provenance, so rewrite details and summary to list what was actually written.
3. The trigger node uses a parent GO term instead of the specific one. GO:0140297 DNA-binding transcription factor binding is the parent of GO:0051059 NF-kappaB binding, which the previous review verified. The PR comment says the parent was chosen because it is already in cache/go/terms.csv. That is not a constraint: just validate-terms checks the cache first and looks up any missing CURIE through OLS, then caches it. Bind GO:0051059 (keep modifier: INCREASED), run just validate-terms so the row is written, and commit cache/go/terms.csv.
🔵 SUGGESTIONS
- The new phenotypes use
category: Immunologic; the rest of the file usesImmunological. Pick one. - The commit rewrote every
>block scalar in the file as a single-quoted string with a trailing blank line ('...\n\n '). That is most of the +432/-369, and it hides the real changes from review. The parsed values are unchanged, so this is not blocking, but please avoid wholesale reformatting in future pushes. - Moriya 2018 (steroid-responsive non-infectious inflammation) remains a good follow-up node once it is cached.
Deep-research cross-check
The Falcon report's main themes are now covered: ectodermal features, combined T/B/innate defect, infections including candidiasis and PCP, GI involvement, lymph node aplasia, poor HSCT outcomes, and the NEMO differential. Remaining gap: non-infectious inflammation (non-blocking, because the source is not cached).
Reference titles
All new evidence items carry reference_title, and the snippet audit checked 56 titles against the cache with 0 mismatches. Every cited paper is about NFKBIA / AD EDA-ID.
just check-genereviews: NO_CHAPTER, so no GeneReviews baseline is required.
|
Addressed the review feedback on
|
|
Claude encountered an error after 1s —— View job I'll analyze this and get back to you. |
|
Claude finished @cmungall's task in 2m 11s —— View job Re-review of PR #14737 (round 3): changes requested, two lines only
Summary checklist
Fixed since round 2: 🟡 Still to fix (two lines)
I made both fixes locally but could not push them: this job's token has no write access to the branch (403). Once these two lines are pushed in one commit, the PR is ready to approve. |
There was a problem hiding this comment.
Re-review (round 3): Ectodermal Dysplasia and Immunodeficiency 2 (PR #14737)
Changes requested, for two lines only. All three round-2 items are fixed in the entry itself. Two leftovers remain in the files around it: the history record still names a wrong HP CURIE, and an enum-cache row is missing.
- Schema validation passes (
just validate) - All PMIDs verified as real (all cached)
- Evidence snippets are exact quotes (48/48 verified)
- Ontology terms correct and specific (the entry binds
GO:0051059NF-kappaB binding) - evidence_source classifications correct
- No pathophysiology bundling
Checks run locally: just validate, just validate-terms, just count-verified-snippets (48/48), just check-causal-targets and just validate-history all pass.
Fixed since round 2:
frequency: FREQUENTis removed from all four new phenotypes.- The trigger node now binds
GO:0051059NF-kappaB binding withINCREASED, and thecache/go/terms.csvrow is committed. - The history record now lists the phenotypes the file actually adds.
- The category spelling is now
Immunologicalthroughout.
🟡 IMPORTANT
1. The history record gives the wrong HP CURIE for the gamma-delta phenotype. The details line says Decreased Gamma-Delta T Cell Proportion (HP:0500255). HP:0500255 is "Increased level of hexanoylglycine in urine" (checked with runoak -i ols:hp info HP:0500255). The entry binds HP:0500271 "Decreased gamma-delta T cell proportion", which is correct. The history record is permanent provenance, so change HP:0500255 to HP:0500271 there.
2. The enum-cache row for GO:0051059 is missing. just validate-terms writes GO:0051059 into cache/enums/molecularfunctionterm_305077eac108.csv between GO:0051018 and GO:0051087, but that file is not in the PR. Validation still passes, because the term is looked up over the network, but CI term validation is meant to run deterministically from committed caches. Run just validate-terms on the entry and commit that one-line change. Do not stage the references_cache/PMID_*.md reformatting the run also produces; it changes nothing.
I made both fixes locally (2 insertions, 1 deletion) but could not push, because this job's token does not have write access to the branch. Please push them in one commit. I will approve once they are in.
🔵 SUGGESTIONS (non-blocking, carried over)
- Moriya 2018 (steroid-responsive non-infectious inflammation) remains a good follow-up node once it is cached.
|
Fixed both remaining items:
|
|
Claude finished @cmungall's task in 1m 43s —— View job Re-review of PR #14737 (round 4): changes requested, one line
Summary checklist
Fixed since round 3: the history record now gives 🟡 Still to fix (one line)
I did not push this fix myself, because the repository's cache rules forbid placing rows by hand. Once it is pushed, the PR is ready to approve. |
There was a problem hiding this comment.
Re-review (round 4): Ectodermal Dysplasia and Immunodeficiency 2 (PR #14737)
Changes requested, for one line. Both round-3 items are fixed, and the entry is ready. The problem is the last commit: it deleted the GO:0006999 row from cache/go/terms.csv, and another entry uses that row.
- Schema validation passes (
just validate) - All PMIDs verified as real (all cached)
- Evidence snippets are exact quotes (48/48 verified)
- Ontology terms correct and specific
- evidence_source classifications correct
- No pathophysiology bundling
Fixed since round 3:
- The history
detailsnow givesHP:0500271for the gamma-delta phenotype.just validate-historypasses. GO:0051059is incache/enums/molecularfunctionterm_305077eac108.csv.
🟡 IMPORTANT
1. The PR deletes a cache row that another entry needs. Compared with origin/main, cache/go/terms.csv adds GO:0051059 and also removes this row:
GO:0006999,nuclear pore organization,2026-10-17T01:02:20.515718
kb/disorders/Nephrotic_Syndrome_Type_12.yaml:169 binds GO:0006999. Without the row, that entry's term validation goes back to the network instead of the committed cache. The row was added on main on 2026-10-17, so its loss is probably a leftover from the merge with main.
To fix it, restore the file from main and let the validator write the new row back. Do not type the row by hand:
git checkout origin/main -- cache/go/terms.csv
just validate-terms kb/disorders/Ectodermal_Dysplasia_and_Immunodeficiency_2.yaml
just normalize-cache
git diff origin/main -- cache/go/terms.csv # expect one added line: GO:0051059Stage only cache/go/terms.csv, as in round 3. I did not push this fix myself because the repository's cache rules forbid hand-placed rows. Once it is pushed, the PR is ready to approve.
🔵 SUGGESTIONS (non-blocking, carried over)
- Moriya 2018 (steroid-responsive non-infectious inflammation) would make a good follow-up node once it is cached.
|
Restored |
|
Claude finished @cmungall's task in 2m 59s —— View job Re-review of PR #14737 (round 5): approved
Summary checklist
Scope: 4 files. The entry, its history record, and one added row in each of the two cache CSVs. No other cache rows change. 🔵 Suggestion (non-blocking): Moriya 2018 (steroid-responsive non-infectious inflammation) would make a good follow-up node once it is cached. |
There was a problem hiding this comment.
Re-review (round 5): approved. The one item left from round 4 is fixed. cache/go/terms.csv now differs from origin/main by exactly one added line, GO:0051059,NF-kappaB binding, and the GO:0006999 row that Nephrotic_Syndrome_Type_12 needs is back.
- Schema validation passes (
just validate) - All PMIDs verified as real (all cached)
- Evidence snippets are exact quotes (48/48 verified)
- Ontology terms correct and specific (the trigger node binds
GO:0051059NF-kappaB binding) - evidence_source classifications correct
- No pathophysiology bundling
Checks run locally: just validate (schema and terms), just count-verified-snippets (48/48), just check-causal-targets, just check-term-cache-integrity and just validate-history-all all pass.
Scope: 4 files: the entry, its history record, and one added row in each of cache/go/terms.csv and cache/enums/molecularfunctionterm_305077eac108.csv. No other cache rows are added or removed.
🔵 Suggestion (non-blocking, carried over): Moriya 2018 (steroid-responsive non-infectious inflammation) would make a good follow-up node once it is cached.
2 participants
Summary
Enriches
kb/disorders/Ectodermal_Dysplasia_and_Immunodeficiency_2.yaml(MONDO:0012714) with structured classifications, inheritance, prevalence, progression, deepenedNFKBIA(IκBα) gain-of-function super-repressor pathophysiology, diagnostic assays, differential diagnoses, and treatment entries.Key Changes
Gain-of-Function NFKBIA (IkappaBalpha) Super-RepressorwithGO:0051059(NF-kappaB binding), modeling impaired Ser32/Ser36 phosphorylation and proteasomal degradation of IκBα, cytoplasmic retention of p50/p65 NF-κB dimers, Ectodysplasin-A receptor (EDAR) signaling failure during ectodermal appendage morphogenesis, and T-cell receptor / TLR-stimulated NF-κB transcriptional failure (PMID:14499265,PMID:29699930).classifications,inheritance(Autosomal Dominant,HP:0000006),prevalence,progression,diagnosis(NFKBIA Sequence Analysis,Lymphocyte Proliferation and NF-kappaB Activation Assay,Immunoglobulin Panel and Vaccine Antibody Titers),differential_diagnoses(Ectodermal Dysplasia and Immunodeficiency 1IKBKG), andtreatments(Immunoglobulin Replacement Therapy,Hematopoietic Stem Cell Transplantation,Antimicrobial Prophylaxis).