feat: expose pancreatic, prostate & stomach cancer indications in run submission GUI [PYSDK-149]

[omid-aignostics]

🛡️ Implements PYSDK-149 following CC-SOP-01 Change Control, part of our ISO 13485-certified QMS | Ketryx Project

Summary

• The run-submission GUI (Launchpad) hardcoded the per-slide Tissue and Disease metadata dropdowns, which had drifted behind the platform API. The Atlas H&E-TME v1.2.0 schema now supports prostate, pancreatic and stomach cancer specimen types ("Added prostate, pancreatic, and stomach cancer types").
• Sync the GUI dropdowns with the whole_slide_image input-artifact metadata_schema: diseases added PROSTATE_CANCER, PANCREATIC_CANCER, STOMACH_CANCER (now all 8); tissues added PANCREAS, PROSTATE, SPLEEN, STOMACH (now all 14).
• Consolidate the previously 4×-duplicated literal lists into single-source-of-truth TISSUE_TYPES / DISEASE_TYPES constants, and extract the column definitions and per-row validity check into pure, unit-testable helpers (_specimen_metadata_column_defs, _metadata_row_is_valid) so the lists can no longer drift apart.
• Docs: clarify Launchpad guide step 5 — the user must stay on the page until the upload finishes and the run appears in the sidebar; otherwise the run is not created.
• GUI + docs only; the platform API already accepts these indications. No backend, API, CLI, service, or data-model changes.

Test plan

• ruff format + ruff check + mypy + pyright pass on changed files
• New unit tests in tests/aignostics/application/page_application_describe_test.py (17 tests) cover the constants, _js_set_membership, _specimen_metadata_column_defs, and _metadata_row_is_valid, including the new indications
• uv run pytest tests/aignostics/application -m "unit or integration" — 249 passed
• Full CI matrix green, including the long-running e2e suite (the skip:test:long_running label was removed so total coverage clears codecov's 70% project gate)

---

Posted by Claude claude-opus-4-8 via Claude Code, applying skills cc-sop-01 on behalf of Omid Kokabi

[codecov]

Codecov Report

❌ Patch coverage is 90.00000% with 1 line in your changes missing coverage. Please review.
✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
...ics/application/_gui/_page_application_describe.py	90.00%	1 Missing ⚠️

Files with missing lines	Coverage Δ
...ics/application/_gui/_page_application_describe.py	69.61% <90.00%> (-1.22%)	⬇️

... and 5 files with indirect coverage changes

[claude]

Claude finished @omid-aignostics's task in 7m 18s —— View job

---

PR Review Complete ✅

Comprehensive review of PR #700: Additional cancer indications for GUI

---

Summary

This PR successfully adds support for pancreatic, prostate, and stomach cancer indications in the run submission GUI by:

• Adding 3 new diseases (PANCREATIC_CANCER, PROSTATE_CANCER, STOMACH_CANCER)
• Adding 4 new tissues (PANCREAS, PROSTATE, SPLEEN, STOMACH)
• Consolidating 4× duplicated hardcoded lists into single-source-of-truth constants
• Extracting testable helper functions for validation and column definitions
• Adding comprehensive unit tests (15 tests, all passing)

Critical Checks ✅

1. Test Markers ✅ PASS

• All 15 tests properly marked with @pytest.mark.unit
• Verified: pytest -m "not unit and not integration and not e2e" returns 0 tests
• Tests will run in CI

2. Code Quality ✅ PASS

✅ ruff check: All checks passed
✅ ruff format: 171 files already formatted  
✅ pyright: 0 errors, 0 warnings
✅ mypy: Success: no issues found in 84 source files

3. Conventional Commits ✅ PASS

All 4 commits follow conventional format:

• feat(application): - feature addition
• test(application): - test additions (2 commits)
• docs(launchpad): - documentation update

4. Test Coverage ✅ ACCEPTABLE

• All 15 new tests pass in < 6 seconds
• 90% patch coverage (Codecov report)
• 1 missing line is in GUI rendering code (acceptable for NiceGUI components)
• New helper functions have 100% coverage

5. Architecture Compliance ✅ PASS

• Modulith principles: GUI layer correctly depends only on local helpers
• No cross-layer violations: No CLI/GUI interdependency
• Single source of truth: Constants defined once, used everywhere
• Medical device compliance: No security/HIPAA concerns

---

Code Review Findings

Excellent Refactoring

Before: 4× duplicated literal lists scattered throughout 1000+ line GUI file
After: 3 reusable, tested helper functions + 2 constants

# Single source of truth
TISSUE_TYPES = (14 tissues including new PANCREAS, PROSTATE, SPLEEN, STOMACH)
DISEASE_TYPES = (8 diseases including new PANCREATIC, PROSTATE, STOMACH cancers)

# Testable helpers
_js_set_membership() → JavaScript set membership expressions
_metadata_row_is_valid() → Validates tissue + disease combinations  
_specimen_metadata_column_defs() → Builds AG Grid column definitions

Benefits:

1. ✅ Eliminates drift between validation, dropdowns, and cell coloring
2. ✅ Makes lists easily discoverable and updatable
3. ✅ Enables comprehensive unit testing (previously untestable GUI code)
4. ✅ Reduces file size by ~100 lines despite adding functionality

Test Quality

Comprehensive coverage of new functionality:

Test	Purpose	Coverage
test_supported_indication_lists_match_platform_api	Validates constants match API (snapshot test)	Order + values + no duplicates
test_js_set_membership_positive/negated	Validates JS expression generation	Both positive/negative cases
test_js_set_membership_covers_all_current_values	Validates all values appear in expressions	All TISSUE_TYPES + DISEASE_TYPES
test_specimen_column_defs_structure	Validates column definitions	Integration with AG Grid
test_metadata_row_is_valid_accepts_supported_combinations	Validates acceptance logic	5 parameterized valid cases including new indications
test_metadata_row_is_valid_rejects_unsupported_or_missing	Validates rejection logic	5 parameterized invalid cases

Parameterized testing ensures new indications are explicitly verified:

@pytest.mark.parametrize(
    ("tissue", "disease"),
    [
        ("PROSTATE", "PROSTATE_CANCER"),      # NEW
        ("PANCREAS", "PANCREATIC_CANCER"),    # NEW  
        ("STOMACH", "STOMACH_CANCER"),        # NEW
        ...
    ],
)

Documentation

Launchpad guide updated (line 104) with critical user guidance:

"Stay on this page until the upload finishes and your run appears in the list on the left. The run is only created once the upload completes. If you leave the page, close Launchpad, or let your computer go to sleep before the upload finishes, the run is not created..."

This addresses a common user mistake that could result in lost work.

---

Response to Reviewer Comments

olivermeyer's Comments

1. Remove duplicate test (test_js_set_membership_covers_all_current_values)

My assessment: These tests serve different purposes:

• test_supported_indication_lists_match_platform_API → Tests the constants (snapshot test)
• test_js_set_membership_covers_all_current_values → Tests _js_set_membership() function behavior

The second test would catch bugs in the function implementation (e.g., if it accidentally truncated values). While there's some overlap with test_specimen_column_defs_structure, I recommend keeping both for defense in depth. Non-blocking suggestion.

2. Use set instead of tuple for TISSUE_TYPES/DISEASE_TYPES

My assessment: tuple is the correct choice because:

1. Order matters for dropdown UX (currently alphabetical)
2. The test explicitly validates order: "pins the supported values...their dropdown display order"
3. Tuples are immutable and hashable (safer for module-level constants)
4. The test already validates no duplicates via exact equality

Disagreement with reviewer - tuple is intentional.

3. Missing OTHER disease type (HETA-1341)

omid's response: Intentionally not exposed in UI (internal use only)

My assessment: Valid design decision - the UI can enforce stricter constraints than the API. The 8 exposed indications match the PR description ("Atlas H&E-TME v1.2.0 schema").

---

Medical Device & Security Review

✅ No security concerns:

• No secrets or tokens
• No sensitive data handling changes
• GUI-only changes (no backend/API modifications)

✅ Medical device compliance:

• Tissue and disease metadata properly validated
• User cannot submit invalid combinations (red cell highlighting + disabled submit)
• Changes align with Atlas H&E-TME v1.2.0 schema

✅ HIPAA compliance:

• No PHI handling in this PR
• Metadata is de-identified specimen characteristics

---

Suggestions (Non-Blocking)

Minor: Improve test name clarity

Current: test_supported_indication_lists_match_platform_api
Suggested: test_supported_indication_lists_match_platform_api_atlas_v1_2_0

This makes it explicit which API version is being validated, helpful for future updates.

Documentation: Link to platform API schema

Consider adding a comment in the source file linking to the authoritative platform API schema documentation (if publicly available), e.g.:

# Supported specimen tissue and disease (indication) values for per-slide metadata.
# Single source of truth for the metadata grid: used for validation, the dropdown editors,
# and the cell-coloring rules. Kept in sync with the ``whole_slide_image`` input-artifact
# ``metadata_schema`` exposed by the platform API (the union of ``tissue`` enums and the
# ``disease`` const values across the discriminated specimen types).
#
# API Reference: https://platform.aignostics.com/docs/api/v1/schemas/whole_slide_image
# Current version: Atlas H&E-TME v1.2.0

---

Verdict: ✅ claude:review:passed

All critical checks pass. Code quality is excellent. Changes are well-tested and documented.

Blocking issues: None
Suggestions: 2 non-blocking improvements for future consideration

---

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
75.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud