Add causal model for Semicircular Canal Dehiscence Syndrome

[kevinschaper]

Summary

• Adds downstream edges on all 4 pathophysiology entries to model the mechanistic cascade: Otic Capsule Bone Biology → Bony Dehiscence → Third Window Effect → Abnormal Sound/Pressure Transmission → vestibular/auditory phenotypes
• Adds sequelae edges on Vertigo (→ Chronic Disequilibrium) and Sound-Induced Nystagmus (→ Oscillopsia)
• Removes creation_date/updated_date fields not yet in the LinkML schema

Causal graph

Otic Capsule Bone Biology
  └─→ Bony Dehiscence of Semicircular Canal
        └─→ Third Window Effect
              ├─→ Abnormal Sound and Pressure Transmission
              │     ├─→ Vertigo ──sequela──→ Chronic Disequilibrium
              │     ├─→ Sound-Induced Nystagmus ──sequela──→ Oscillopsia
              │     └─→ Chronic Disequilibrium
              ├─→ Conductive Hearing Loss
              ├─→ Autophony
              ├─→ Hyperacusis
              └─→ Pulsatile Tinnitus

Test plan

• just validate passes
• HTML page regenerated

🤖 Generated with Claude Code

[claude]

Claude finished @kevinschaper's task in 1m 46s —— View job

---

PR Review: Add causal model for Semicircular Canal Dehiscence Syndrome

Tasks:

• Gather context
• Read changed YAML file
• Review schema for downstream and sequelae fields
• Assess biological/clinical validity
• Check conformance with guidelines
• Post feedback

---

Summary

This PR adds a well-structured causal graph to the SCDS entry using downstream and sequelae edges, and removes creation_date/updated_date fields that aren't yet in the schema. Overall the biology is sound and the modeling is coherent. A few items worth addressing:

---

Issues

1. Mismatch between preferred_term and GO term label (pre-existing, Third Window Effect node)

In the Third Window Effect pathophysiology entry, the biological process annotation reads:

- preferred_term: mechanotransduction
  term:
    id: GO:0050957
    label: sensory perception of sound

GO:0050957 is "sensory perception of sound", not "mechanotransduction" (GO:0009612 or GO:0051899). This is a pre-existing issue (not introduced by this PR), but since the PR touches this node it's worth flagging. The preferred_term and term.label should be consistent. Fix this →

2. downstream targets from pathophysiology → phenotypes cross schema boundary without causal_link_type

The Third Window Effect node lists downstream targets that are phenotypes rather than pathophysiology nodes: Conductive Hearing Loss, Autophony, Hyperacusis, Pulsatile Tinnitus. Similarly Abnormal Sound and Pressure Transmission → Vertigo, Sound-Induced Nystagmus, Chronic Disequilibrium.

The schema supports causal_link_type values (DIRECT, INDIRECT_KNOWN_INTERMEDIATES, INDIRECT_UNKNOWN_INTERMEDIATES, UNKNOWN) and they are left unset here. Other entries in the KB (e.g., Asthma, Marfan Syndrome) also omit this field, so it appears to be optional — but specifying it would be valuable since the relationship from a pathophysiology mechanism to a clinical phenotype is often DIRECT (or INDIRECT_KNOWN_INTERMEDIATES where cellular/molecular steps are omitted). Consider annotating at least the more distal edges.

3. Redundant Chronic Disequilibrium path

In the causal graph, Chronic Disequilibrium appears as a downstream target of Abnormal Sound and Pressure Transmission (as a pathophysiology consequence) and as a sequela of Vertigo (phenotype-level). Both are biologically defensible:

• The pathophysiology path models persistent low-level vestibular stimulation causing chronic imbalance directly.
• The phenotype sequela path models recurrent vertigo → secondary chronic disequilibrium.

These are mechanistically distinct pathways (direct vs. indirect via episodic vertigo), so duplicating the target is intentional and correct. The descriptions differentiate them well. No change needed, but it's worth a brief comment in the notes field explaining that chronic disequilibrium has both direct (continuous stimulation) and indirect (post-vertigo) contributions.

---

Minor Observations

• Graph ordering: The pathophysiology entries are listed in this order: Bony Dehiscence, Third Window Effect, Abnormal Sound and Pressure Transmission, Otic Capsule Bone Biology. Logically the cascade starts with Otic Capsule Bone Biology. Reordering to match the mechanistic flow (Otic Capsule → Bony Dehiscence → Third Window → Abnormal Transmission) would improve readability, though this has no functional impact.

• Pulsatile tinnitus mechanism: The downstream edge from Third Window Effect → Pulsatile Tinnitus describes vascular pulsations transmitted via the third window. This is correct but slightly simplified — the primary mechanism is enhanced bone conduction of vascular pulsations through the dehiscence (same pathway as autophony). The description is acceptable as written.

• creation_date/updated_date removal: Correct to remove these until the schema supports them. The PR description confirms they are not yet in the LinkML schema.

---

Biological/Clinical Validity

The causal cascade is mechanistically accurate and well-supported by the cited literature:

• PMID:32982922 (biomechanics review) is the primary source for the third window mechanism — appropriate.
• PMID:22312921 (histopathology case) supporting the structural basis of Otic Capsule Biology → Bony Dehiscence is apt.
• PMID:16222184 (Minor et al. original clinical series) appropriately anchors the clinical phenotype frequencies.
• The sequela modeling (Sound-Induced Nystagmus → Oscillopsia) correctly captures that nystagmus from vestibular stimulation produces the perception of visual oscillation.

---

Verdict

The causal model is biologically sound and represents a meaningful enhancement to the SCDS entry. The pre-existing preferred_term/GO label mismatch in Third Window Effect should be corrected (it would likely fail term validation). All other items are minor suggestions.

[claude]

Claude finished @kevinschaper's task in 1m 37s —— View job

---

PR Review: Add causal model for Semicircular Canal Dehiscence Syndrome

Tasks:

• Gather context
• Read changed YAML file
• Check git diff for changes since previous review
• Review biological/clinical validity
• Check conformance with guidelines
• Post feedback

---

Summary

This PR adds a well-structured causal graph to the SCDS entry using downstream and sequelae edges, and correctly removes creation_date/updated_date fields that are not yet in the LinkML schema. The biology is accurate and the mechanistic cascade is clearly modeled. One pre-existing issue (GO term label mismatch) remains unaddressed and should be fixed before merge.

---

Issues

1. Pre-existing GO term label mismatch in Third Window Effect node (should be fixed)

The Third Window Effect pathophysiology entry (lines 53–57) reads:

biological_processes:
- preferred_term: mechanotransduction
  term:
    id: GO:0050957
    label: sensory perception of sound

GO:0050957 has the canonical label "sensory perception of sound" — not "mechanotransduction". The preferred_term and term.label must be consistent with the ontology's canonical label. This will likely fail just validate-terms. This is a pre-existing bug, not introduced by this PR, but the PR touches this node so it's a good time to fix it.

Two options:

• Option A: Keep GO:0050957 but set preferred_term: sensory perception of sound (or a human-friendly alias).
• Option B: Find the correct GO term for mechanotransduction (GO:0009612 — "response to mechanical stimulus" or similar) and replace the ID.

Fix this →

---

Observations and Suggestions (Non-blocking)

2. causal_link_type left unset on all edges

CausalEdge supports a causal_link_type field (DIRECT, INDIRECT_KNOWN_INTERMEDIATES, INDIRECT_UNKNOWN_INTERMEDIATES, UNKNOWN) which is optional in the schema. None of the new edges populate it. While not required, several of these relationships are clearly DIRECT (e.g., Abnormal Sound and Pressure Transmission → Vertigo — no unknown intermediates) and annotating them would improve the semantic value of the graph. This is a suggestion, not a blocker.

3. Redundant Chronic Disequilibrium as downstream target

Chronic Disequilibrium appears as:

• A downstream target of Abnormal Sound and Pressure Transmission (persistent low-level stimulation pathway)
• A sequela of Vertigo phenotype (recurrent episodic pathway)

Both are biologically correct — they model distinct mechanisms. The notes field could briefly clarify this dual contribution to avoid confusion for future curators.

4. Pathophysiology ordering

The pathophysiology list order is: Bony Dehiscence → Third Window Effect → Abnormal Sound/Pressure Transmission → Otic Capsule Bone Biology. Logically the cascade begins with Otic Capsule Bone Biology. Reordering to match mechanistic flow would improve readability, though this has no functional impact.

---

Biological/Clinical Validity

The causal cascade is mechanistically accurate:

Edge	Assessment
Otic Capsule Bone Biology → Bony Dehiscence	Correct — thin otic capsule is the structural predisposition
Bony Dehiscence → Third Window Effect	Correct — the dehiscence is the third window
Third Window Effect → Abnormal Sound/Pressure Transmission	Correct — low-impedance pathway aberrantly transmits energy
Abnormal Sound/Pressure → Vertigo, Nystagmus	Correct — cupula deflection via Tullio/Hennebert mechanisms
Sound-Induced Nystagmus → Oscillopsia (sequela)	Correct — nystagmus causes visual oscillation perception
Vertigo → Chronic Disequilibrium (sequela)	Correct — recurrent vertigo contributes to persistent imbalance
Third Window → Conductive Hearing Loss, Autophony, Hyperacusis, Pulsatile Tinnitus	All correct via the enhanced bone conduction / acoustic shunting mechanism

Evidence citations (PMID:32982922, PMID:22312921, PMID:16222184, PMID:18223508) are appropriate and well-matched to the claims.

---

Schema Conformance

• downstream and sequelae are valid schema fields on pathophysiology and phenotype nodes respectively. ✓
• creation_date/updated_date removal is correct — these fields are not in the current schema. ✓
• target (required field in CausalEdge) is populated for all edges. ✓
• All targets reference names that exist elsewhere in the file (phenotypes or pathophysiology entries). ✓

---

Verdict

Approve after fixing the GO term mismatch. The causal model is biologically sound, well-structured, and a meaningful addition to the SCDS entry. The GO label mismatch pre-dates this PR but should be corrected here since the node is being modified.

[cmungall]

@dragon-ai-agent please make an issue - the "fix this" buttons dont work as expected e.g. https://claude.ai/code?q=In%20kb%2Fdisorders%2FSemicircular_Canal_Dehiscence_Syndrome.yaml%20in%20the%20%22Third%20Window%20Effect%22%20pathophysiology%20entry%2C%20the%20biological_processes%20annotation%20has%20preferred_term%3A%20mechanotransduction%20but%20the%20GO%20term%20is%20GO%3A0050957%20(sensory%20perception%20of%20sound).%20Either%20change%20preferred_term%20to%20match%20the%20actual%20label%2C%20or%20replace%20with%20the%20correct%20GO%20term%20for%20mechanotransduction.&repo=monarch-initiative/dismech ==> this takes me to a prompt:

In kb/disorders/Houge-Janssens_Syndrome.yaml at line 195-199, replace the biological process term GO:0006470 (protein dephosphorylation) with the molecular function term GO:0004722 (protein serine/treonine phosphatase activity). Update both the preferred_term and label to match.

this has nothing to do with the current PR! This may be a chrome thing and not a you thing but still it's confusing. Also cross-reference with issues about making targeted changes in PRs that can easily be committed. This would improve flows a lot.

[github-actions]

I've created issue #392 to track this: #392

The issue documents that the "Fix this" compliance dashboard buttons redirect to https://claude.ai/code?q=... in the browser rather than performing (or initiating) the fix locally. The issue includes a concrete example from the Semicircular Canal Dehiscence Syndrome case you linked, steps to reproduce, and suggested directions for a fix.

[cmungall]

@kevinschaper can you have your agent address the one remaining thing we want to be precise about mechanotransduction (or not have a term)