Data Analysis

wgcna-analysis

Identify co-expression gene modules and correlate them with clinical traits using Weighted Gene Co-expression Network Analysis. Inputs: expression matrix, trait/phenotype table. Outputs: module color assignments, trait correlation heatmap, hub gene list per module.

90100Total Score
Core Capability
91 / 100
Functional Suitability
11 / 12
Reliability
10 / 12
Performance & Context
7 / 8
Agent Usability
15 / 16
Human Usability
7 / 8
Security
11 / 12
Maintainability
11 / 12
Agent-Specific
19 / 20
Medical Task
25 / 25 Passed
90Updated smoke test
5/5
88Signed bicor run
5/5
91Documentation conformance
5/5
91Explicit module export
5/5
89Chunked loading path
5/5

Veto GatesRequired pass for any deployment consideration

Skill Veto✓ All 4 gates passed
✓
Operational Stability
System remains stable across varied inputs and edge cases
PASS
✓
Structural Consistency
Output structure conforms to expected skill contract format
PASS
✓
Result Determinism
Equivalent inputs produce semantically equivalent outputs
PASS
✓
System Security
No prompt injection, data leakage, or unsafe tool use detected
PASS
Research Veto✅ PASS — Applicable
DimensionResultDetail
Scientific IntegrityPASSNo fabricated statistics, identifiers, or claims were introduced during any audited execution.
Practice BoundariesPASSThe skill remained inside computational analysis boundaries and did not emit diagnostic or prescriptive advice.
Methodological GroundPASSAll audited runs used coherent WGCNA methods with documented parameter branches and no methodological redline violation.
Code UsabilityPASSThe R pipeline executed successfully for all five audited inputs, including the repaired documentation-conformant smoke test.

Core Capability91 / 100 — 8 Categories

Functional Suitability
The core WGCNA use cases are covered well and the smoke test now matches the shipped fixture set.
11 / 12
92%
Reliability
Error handling is strong and actionable, though failed runs can still leave partially populated output directories behind.
10 / 12
83%
Performance & Context
Progressive disclosure is strong; minor context bloat remains in the auxiliary CLI examples.
7 / 8
88%
Agent Usability
The documentation is clearer after the out-of-scope and verification updates; only minor wording redundancy remains.
15 / 16
94%
Human Usability
Trigger language is natural for bioinformatics users and stop conditions are now easier to apply.
7 / 8
88%
Security
Input validation is solid and no dangerous execution primitives were found; data-retention boundaries could still be stated more explicitly.
11 / 12
92%
Maintainability
The scripts remain modular, but the bundled validator still does not assert every documented option-dependent artifact.
11 / 12
92%
Agent-Specific
Trigger precision, layering, and escape hatches are strong after the revision.
19 / 20
95%
Core Capability Total91 / 100

Medical TaskExecution Average: 89.8 / 100 — Assertions: 25/25 Passed

90
Canonical
Updated smoke test
5/5 ✓
88
Variant A
Signed bicor run
5/5 ✓
91
Edge
Documentation conformance
5/5 ✓
91
Variant B
Explicit module export
5/5 ✓
89
Stress
Chunked loading path
5/5 ✓
90
Canonical✅ Pass
Updated smoke test

Executed successfully and passed the bundled baseline validator.

Basic 37/40|Specialized 53/60|Total 90/100
✅A1The documented smoke-test command completes successfully.
✅A2The baseline validator passes on the smoke-test output directory.
✅A3The run exports a ranked module summary and a per-module gene table.
✅A4The workflow stays within its stated scope and does not fabricate scientific claims.
✅A5The analysis completes without manual intervention.
Pass rate: 5 / 5
88
Variant A✅ Pass
Signed bicor run

Alternative network settings executed successfully and passed validation.

Basic 36/40|Specialized 52/60|Total 88/100
✅A1The alternative signed-network parameter set executes successfully.
✅A2The baseline output bundle remains intact under a parameter variation.
✅A3The skill accepts documented correlation and network-type switches.
✅A4The run produces a selected module export in the stated format.
✅A5No fabricated scientific claims or unsafe instructions appear in the output.
Pass rate: 5 / 5
91
Edge✅ Pass
Documentation conformance

The repaired smoke-test documentation is now fully aligned with the bundled assets.

Basic 38/40|Specialized 53/60|Total 91/100
✅A1The repository no longer references the removed tests/data/expression_subset.csv fixture.
✅A2The documented smoke-test command completes successfully.
✅A3The validation command succeeds against the documented output directory.
✅A4The repaired documentation is recoverable and self-consistent.
✅A5The script does not perform unsafe operations during documentation-conformant execution.
Pass rate: 5 / 5
91
Variant B✅ Pass
Explicit module export

Explicit trait and multi-module export completed successfully.

Basic 37/40|Specialized 54/60|Total 91/100
✅A1The skill accepts an explicit trait selection within the encoded trait matrix.
✅A2Two requested modules are exported when they are valid.
✅A3Module-specific scatter plots are produced for the requested exports.
✅A4The module-selection behavior matches the stated CLI contract.
✅A5The output remains within the intended WGCNA analysis scope.
Pass rate: 5 / 5
89
Stress✅ Pass
Chunked loading path

Chunked loading completed successfully and repeated smoke-test outputs remained deterministic.

Basic 36/40|Specialized 53/60|Total 89/100
✅A1The chunked-loading execution path completes successfully.
✅A2The chunked workflow emits useful progress feedback.
✅A3The chunked run still produces the baseline required outputs.
✅A4Repeated runs with the same seed remain deterministic on key summary files.
✅A5The stress run remains within the intended WGCNA workflow.
Pass rate: 5 / 5
Medical Task Total89.8 / 100

Key Strengths

  • The documented smoke test now matches the shipped fixtures and executes successfully end to end.
  • The workflow is reproducible: repeated smoke-test runs produced identical hashes for key summary outputs.
  • Input validation and troubleshooting guidance are strong, with consistent SKILL_* error codes and actionable recovery advice.
  • The implementation is modular and supports both standard and chunked loading paths without changing the canonical summary on bundled data.