Tooluniverse Multiomic Disease Characterization

Name: Tooluniverse Multiomic Disease Characterization
Author: mims-harvard

mims-harvard/tooluniverse

315 installs
1.6k repo stars
Updated August 4, 2026
mims-harvard/tooluniverse

tooluniverse-multiomic-disease-characterization is a biomedical agent skill that integrates genomics, transcriptomics, proteomics, and pathway data across nine phases for developers producing systems-level disease mechan

About

tooluniverse-multiomic-disease-characterization is a mims-harvard/ToolUniverse skill for full-omics disease deep dives. After disease disambiguation to MONDO/EFO identifiers, it runs nine phases covering GWAS genetics, GTEx expression across 54 tissues, STRING PPI networks, Reactome and Enrichr pathway enrichment, GO annotation, therapeutic landscape via DGIdb and clinical trials, cross-layer integration, and a 0–100 Multi-Omics Confidence Score with T1–T4 evidence grading. The report-first workflow creates a markdown report before populating sections progressively. Reference files include tool-reference.md, report-template.md, integration-scoring.md, response-formats.md, and use-patterns.md. Use when agents must connect variants to expression, protein interactions, pathways, druggable targets, and biomarker hypotheses in one structured translational report.

Cross-omic disease integration tools
Biomarker and subtype hypothesis support
Layered evidence aggregation
ToolUniverse multi-omic connectors
Translational research agent workflows

Tooluniverse Multiomic Disease Characterization by the numbers

315 all-time installs (skills.sh)
+5 installs in the week ending Aug 4, 2026 (Skillselion tracking)
Ranked #575 of 2,064 Data Science & ML skills by installs in the Skillselion catalog
Data as of Aug 5, 2026 (Skillselion catalog sync)

npx skills add https://github.com/mims-harvard/tooluniverse --skill tooluniverse-multiomic-disease-characterization

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

[![Listed on Skillselion](https://skillselion.com/badge/skills/mims-harvard/tooluniverse/tooluniverse-multiomic-disease-characterization.svg)](https://skillselion.com/skills/mims-harvard/tooluniverse/tooluniverse-multiomic-disease-characterization)

Installs	315
repo stars	★ 1.6k
Last updated	August 4, 2026
Repository	mims-harvard/tooluniverse ↗

How do you characterize disease mechanisms across omics layers?

Help agents integrate genomics, transcriptomics, proteomics, and metabolomics signals to characterize disease mechanisms, biomarkers, and subtype hypotheses in translational projects.

Who is it for?

Translational bioinformatics engineers building agents that synthesize genomics, transcriptomics, proteomics, and pathway evidence into unified disease characterization reports.

Skip if: Single-gene drug-target validation only, adverse-event safety profiling, or quick disease summaries without cross-omics integration and scored evidence grading.

When should I use this skill?

User needs multi-omics disease characterization, cross-layer mechanism mapping, biomarker discovery, or druggable target identification from integrated molecular data

What you get

Multi-omics markdown reports, cross-layer gene concordance tables, biomarker candidates, druggable targets, and confidence scores

multi-omics disease reports
biomarker candidate lists
druggable target rankings

By the numbers

Runs a 9-phase multi-omics characterization pipeline
GTEx expression summary covers 54 tissues
Multi-Omics Confidence Score ranges 0–100 with T1–T4 evidence grades

Files

SKILL.mdMarkdownGitHub ↗

Multi-Omics Disease Characterization Pipeline

Characterize diseases across multiple molecular layers (genomics, transcriptomics, proteomics, pathways) to provide systems-level understanding of disease mechanisms, identify therapeutic opportunities, and discover biomarker candidates.

KEY PRINCIPLES: 1. Report-first approach - Create report file FIRST, then populate progressively 2. Disease disambiguation FIRST - Resolve all identifiers before omics analysis 3. Layer-by-layer analysis - Systematically cover all omics layers 4. Cross-layer integration - Identify genes/targets appearing in multiple layers 5. Evidence grading - Grade all evidence as T1 (human/clinical) to T4 (computational) 6. Tissue context - Emphasize disease-relevant tissues/organs 7. Quantitative scoring - Multi-Omics Confidence Score (0-100) 8. Druggable focus - Prioritize targets with therapeutic potential 9. Biomarker identification - Highlight diagnostic/prognostic markers 10. Mechanistic synthesis - Generate testable hypotheses 11. Source references - Every statement must cite tool/database 12. Completeness checklist - Mandatory section showing analysis coverage 13. English-first queries - Always use English terms in tool calls. Respond in user's language

Multi-omics disease characterization asks: what molecular layers are dysregulated? Genomic mutations → transcriptomic changes → proteomic effects → metabolomic consequences. Concordance across layers strengthens the finding. Discordance reveals regulatory complexity.

LOOK UP, DON'T GUESS

When uncertain about any scientific fact, SEARCH databases first rather than reasoning from memory. A database-verified answer is always more reliable than a guess.

---

COMPUTE, DON'T DESCRIBE

When analysis requires computation (statistics, data processing, scoring, enrichment), write and run Python code via Bash. Don't describe what you would do — execute it and report actual results. Use ToolUniverse tools to retrieve data, then Python (pandas, scipy, statsmodels, matplotlib) to analyze it.

When to Use This Skill

Apply when users:

Ask about disease mechanisms across omics layers
Need multi-omics characterization of a disease
Want to understand disease at the systems biology level
Ask "What pathways/genes/proteins are involved in [disease]?"
Need biomarker discovery for a disease
Want to identify druggable targets from disease profiling
Ask for integrated genomics + transcriptomics + proteomics analysis
Need cross-layer concordance analysis
Ask about disease network biology / hub genes

NOT for (use other skills instead):

Single gene/target validation -> Use tooluniverse-drug-target-validation
Drug safety profiling -> Use tooluniverse-adverse-event-detection
General disease overview -> Use tooluniverse-disease-research
Variant interpretation -> Use tooluniverse-variant-interpretation
GWAS-specific analysis -> Use tooluniverse-gwas-* skills
Pathway-only analysis -> Use tooluniverse-systems-biology

---

Input Parameters

Parameter	Required	Description	Example
disease	Yes	Disease name, OMIM ID, EFO ID, or MONDO ID	`Alzheimer disease`, `MONDO_0004975`
tissue	No	Tissue/organ of interest	`brain`, `liver`, `blood`
focus_layers	No	Specific omics layers to emphasize	`genomics`, `transcriptomics`, `pathways`

---

Pipeline Overview

The pipeline runs 9 phases sequentially. Each phase uses specific tools documented in detail in tool-reference.md.

Phase 0: Disease Disambiguation (ALWAYS FIRST)

Resolve disease to standard identifiers (MONDO/EFO) for all downstream queries.

Primary tool: OpenTargets_get_disease_id_description_by_name
Get description, synonyms, therapeutic areas, disease hierarchy, cross-references
CRITICAL: Disease IDs use underscore format (e.g., MONDO_0004975), NOT colon
If ambiguous, present top 3-5 options and ask user to select

Phase 1: Genomics Layer

Identify genetic variants, GWAS associations, and genetically implicated genes.

Tools: gwas_search_associations (use efo_id for precision, not free-text disease_trait), gwas_get_snps_for_gene, ClinVar, OpenTargets associated targets
gnomad_get_gene_constraints — gene constraint metrics (pLI, oe_lof) to interpret whether LoF variants are tolerated vs. haploinsufficient
Get top 10-15 genes with genetic evidence scores; track Ensembl IDs for downstream phases

Phase 2: Transcriptomics Layer

Identify differentially expressed genes, tissue-specific expression, and expression-based biomarkers.

GTEx_get_expression_summary — baseline expression across 54 tissues (accepts gene_symbol directly)
Tools: Expression Atlas, HPA (tissue expression), EuropePMC scores
Check expression in disease-relevant tissues for top genes from Phase 1

Phase 3: Proteomics & Interaction Layer

Map protein-protein interactions, identify hub genes, and characterize interaction networks.

UniProt_get_function_by_accession — protein function narrative (essential for mechanistic context)
Tools: STRING_get_network (param: identifiers, species=9606), intact_get_interactions, HumanBase
Build PPI network from top 15-20 genes; identify hub genes by degree centrality

Phase 4: Pathway & Network Layer

Identify enriched biological pathways and cross-pathway connections.

ReactomeAnalysis_pathway_enrichment — identifiers are newline-separated (\n), NOT space-separated
enrichr_gene_enrichment_analysis — param: gene_list (array), libs (array). NOTE: data field is a JSON string that needs parsing
kegg_search_pathway — pathway keyword search

Phase 5: Gene Ontology & Functional Annotation

Characterize biological processes, molecular functions, and cellular components.

Tools: Enrichr (GO libraries), QuickGO, GO annotations, OpenTargets GO
Run GO enrichment for all 3 aspects (BP, MF, CC)

Phase 6: Therapeutic Landscape

Map approved drugs, druggable targets, repurposing opportunities, and clinical trials.

DGIdb_get_drug_gene_interactions — drug interactions by gene (param: genes as array). Often more comprehensive than OpenTargets for drug-gene data.
OpenTargets drugs/tractability (use EFO IDs like EFO_0000384 for Crohn's, not MONDO — MONDO IDs may return null for drug queries)
search_clinical_trials — query_term is REQUIRED

Phase 7: Multi-Omics Integration

Integrate findings across all layers. See integration-scoring.md for full details.

Cross-layer gene concordance: count layers per gene, score multi-layer hub genes
Direction concordance: genetics + expression agreement
Biomarker identification: diagnostic, prognostic, predictive
Mechanistic hypothesis generation

Phase 8: Report Finalization

Write executive summary, calculate confidence score, verify completeness.

See integration-scoring.md for quality checklist and scoring formula

---

Key Tool Parameter Notes

These are the most common parameter pitfalls:

OpenTargets disease IDs: underscore format (MONDO_0004975), NOT colon
STRING protein_ids: must be array (['APOE']), not string
enrichr libs: must be array (['KEGG_2021_Human'])
HPA_get_rna_expression_by_source: ALL 3 params required (gene_name, source_type, source_name)
humanbase_ppi_analysis: ALL params required (gene_list, tissue, max_node, interaction, string_mode)
expression_atlas_disease_target_score: pageSize is REQUIRED
search_clinical_trials: query_term is REQUIRED even if condition is provided

For full tool parameters and per-phase workflows, see tool-reference.md.

---

Reference Files

All detailed content is in reference files in this directory:

File	Contents
`tool-reference.md`	Full tool parameters, inputs/outputs, per-phase workflows, quick reference table
`report-template.md`	Complete report markdown template with all sections and checklists
`integration-scoring.md`	Confidence score formula (0-100), evidence grading (T1-T4), integration procedures, quality checklist
`response-formats.md`	Verified JSON response structures for key tools
`use-patterns.md`	Common use patterns, edge case handling, fallback strategies

# API Keys for ToolUniverse
# Copy this file to .env and fill in your actual API keys

BIOGRID_API_KEY=your_api_key_here

BOLTZ_MCP_SERVER_HOST=your_api_key_here

BRENDA_EMAIL=your_api_key_here

BRENDA_PASSWORD=your_api_key_here

DISGENET_API_KEY=your_api_key_here

EXPERT_FEEDBACK_MCP_SERVER_URL=your_api_key_here

NVIDIA_API_KEY=your_api_key_here

OMIM_API_KEY=your_api_key_here

TXAGENT_MCP_SERVER_HOST=your_api_key_here

USPTO_API_KEY=your_api_key_here

USPTO_MCP_SERVER_HOST=your_api_key_here

Multi-Omics Integration & Confidence Scoring

---

Multi-Omics Confidence Score (0-100)

Score Components

Data Availability (0-40 points):

Genomics data available (GWAS or rare variants): 10 points
Transcriptomics data available (DEGs or expression): 10 points
Protein data available (PPI or expression): 5 points
Pathway data available (enriched pathways): 10 points
Clinical/drug data available (approved drugs or trials): 5 points

Evidence Concordance (0-40 points):

Multi-layer genes (appear in 3+ layers): up to 20 points (2 per gene, max 10 genes)
Consistent direction (genetics + expression concordant): 10 points
Pathway-gene concordance (genes found in enriched pathways): 10 points

Evidence Quality (0-20 points):

Strong genetic evidence (GWAS p < 5e-8): 10 points
Clinical validation (approved drugs): 10 points

Score Interpretation

Score	Tier	Interpretation
80-100	Excellent	Comprehensive multi-omics coverage, high confidence, strong cross-layer concordance
60-79	Good	Good coverage across most layers, some gaps
40-59	Moderate	Moderate coverage, limited cross-layer integration
0-39	Limited	Limited data, single-layer analysis dominates

Evidence Grading System

Tier	Symbol	Criteria	Examples
T1	[T1]	Direct human evidence, clinical proof	FDA-approved drug, GWAS hit (p<5e-8), clinical trial result
T2	[T2]	Experimental evidence	Differential expression (validated), functional screen, mouse KO
T3	[T3]	Computational/database evidence	PPI network, pathway mapping, expression correlation
T4	[T4]	Annotation/prediction only	GO annotation, text-mined association, predicted interaction

---

Phase 7: Multi-Omics Integration

Objective: Integrate findings across all layers to identify cross-layer genes, calculate concordance, and generate mechanistic hypotheses.

Cross-Layer Gene Concordance Analysis

For each gene found in the analysis:

1. Count layers: In how many omics layers does this gene appear?

Genomics (GWAS, rare variants, genetic association)
Transcriptomics (DEGs, expression score)
Proteomics (PPI hub, protein expression)
Pathways (enriched pathway member)
Therapeutics (drug target)

2. Score genes: Genes appearing in 3+ layers are "multi-omics hub genes"

3. Direction concordance: Do genetics and expression agree?

Risk allele + upregulated = concordant gain-of-function
Risk allele + downregulated = concordant loss-of-function
Discordant = needs investigation

Biomarker Identification

For each multi-omics hub gene, assess biomarker potential:

Diagnostic: Gene expression distinguishes disease vs healthy
Prognostic: Expression/variant predicts outcome (cancer prognostics from HPA)
Predictive: Variant/expression predicts treatment response (pharmacogenomics)
Evidence level: Number of supporting omics layers

Mechanistic Hypothesis Generation

From the integrated data: 1. Identify the most supported biological processes (GO + pathways) 2. Map causal chain: genetic variant -> gene expression -> protein function -> pathway disruption -> disease 3. Identify intervention points (druggable nodes in the causal chain) 4. Generate testable hypotheses

Confidence Score Calculation

Calculate the Multi-Omics Confidence Score (0-100) based on:

Data availability across layers
Cross-layer concordance
Evidence quality
Clinical validation

---

Phase 8: Report Finalization

Executive Summary

Write a 2-3 sentence synthesis covering:

Disease mechanism in systems terms
Key genes/pathways identified
Therapeutic opportunities

Final Report Quality Checklist

Before presenting to user, verify:

[ ] All 8 sections have content (or marked as "No data available")
[ ] Every data point has a source citation
[ ] Executive summary reflects key findings
[ ] Multi-Omics Confidence Score calculated
[ ] Top 20 genes ranked by multi-omics evidence
[ ] Top 10 enriched pathways listed
[ ] Biomarker candidates identified
[ ] Cross-layer concordance table complete
[ ] Therapeutic opportunities summarized
[ ] Mechanistic hypotheses generated
[ ] Data Availability Checklist complete
[ ] Completeness Checklist complete
[ ] References section lists all tools used

Multi-Omics Disease Characterization Report Template

Create this file at the start: {disease_name}_multiomic_report.md

# Multi-Omics Disease Characterization: {Disease Name}

**Report Generated**: {date}
**Disease Identifiers**: (to be filled)
**Multi-Omics Confidence Score**: (to be calculated)

---

## Executive Summary

(2-3 sentence disease mechanism synthesis - fill after all layers complete)

---

## 1. Disease Definition & Context

### Disease Identifiers
| System | ID | Source |
|--------|-----|--------|

### Description
### Synonyms
### Disease Hierarchy (parents/children)
### Affected Tissues/Organs
### Therapeutic Areas

**Sources**: (tools used)

---

## 2. Genomics Layer

### 2.1 GWAS Associations
| SNP | P-value | Effect | Gene | Study | Source |
|-----|---------|--------|------|-------|--------|

### 2.2 GWAS Studies Summary
| Study ID | Trait | Sample Size | Year | Source |
|----------|-------|-------------|------|--------|

### 2.3 Associated Genes (Genetic Evidence)
| Gene | Ensembl ID | Association Score | Evidence Type | Source |
|------|------------|-------------------|---------------|--------|

### 2.4 Rare Variants (ClinVar)
| Variant | Gene | Clinical Significance | Source |
|---------|------|-----------------------|--------|

### Genomics Layer Summary
- Total GWAS hits:
- Top genes by genetic evidence:
- Genetic architecture:

**Sources**: (tools used)

---

## 3. Transcriptomics Layer

### 3.1 Differential Expression Studies
| Experiment | Condition | Up-regulated | Down-regulated | Source |
|------------|-----------|--------------|----------------|--------|

### 3.2 Expression Atlas Disease Evidence
| Gene | Score | Source |
|------|-------|--------|

### 3.3 Tissue Expression Patterns (GTEx/HPA)
| Gene | Tissue | Expression Level | Source |
|------|--------|-----------------|--------|

### 3.4 Biomarker Candidates (Expression-Based)
| Gene | Tissue Specificity | Fold Change | Evidence | Source |
|------|-------------------|-------------|----------|--------|

### Transcriptomics Layer Summary
- Differential expression datasets:
- Top DEGs:
- Tissue-specific patterns:

**Sources**: (tools used)

---

## 4. Proteomics & Interaction Layer

### 4.1 Protein-Protein Interactions (STRING)
| Protein A | Protein B | Score | Source |
|-----------|-----------|-------|--------|

### 4.2 Hub Genes (Network Centrality)
| Gene | Degree | Betweenness | Role | Source |
|------|--------|-------------|------|--------|

### 4.3 Protein Complexes (IntAct)
| Complex | Members | Function | Source |
|---------|---------|----------|--------|

### 4.4 Tissue-Specific PPI Network
| Gene | Interaction Score | Tissue | Source |
|------|-------------------|--------|--------|

### Proteomics Layer Summary
- Total PPIs:
- Hub genes:
- Network modules:

**Sources**: (tools used)

---

## 5. Pathway & Network Layer

### 5.1 Enriched Pathways (Enrichr/Reactome)
| Pathway | Database | P-value | Genes | Source |
|---------|----------|---------|-------|--------|

### 5.2 Reactome Pathway Details
| Pathway ID | Name | Genes Involved | Source |
|------------|------|----------------|--------|

### 5.3 KEGG Pathways
| Pathway ID | Name | Description | Source |
|------------|------|-------------|--------|

### 5.4 WikiPathways
| Pathway ID | Name | Organism | Source |
|------------|------|----------|--------|

### Pathway Layer Summary
- Top enriched pathways:
- Key pathway nodes:
- Cross-pathway connections:

**Sources**: (tools used)

---

## 6. Gene Ontology & Functional Annotation

### 6.1 Biological Processes
| GO Term | Name | P-value | Genes | Source |
|---------|------|---------|-------|--------|

### 6.2 Molecular Functions
| GO Term | Name | P-value | Genes | Source |
|---------|------|---------|-------|--------|

### 6.3 Cellular Components
| GO Term | Name | P-value | Genes | Source |
|---------|------|---------|-------|--------|

**Sources**: (tools used)

---

## 7. Therapeutic Landscape

### 7.1 Approved Drugs
| Drug | ChEMBL ID | Mechanism | Target | Phase | Source |
|------|-----------|-----------|--------|-------|--------|

### 7.2 Druggable Targets
| Gene | Tractability | Modality | Clinical Precedent | Source |
|------|-------------|----------|-------------------|--------|

### 7.3 Drug Repurposing Candidates
| Drug | Original Indication | Mechanism | Target | Source |
|------|---------------------|-----------|--------|--------|

### 7.4 Clinical Trials
| NCT ID | Title | Phase | Status | Intervention | Source |
|--------|-------|-------|--------|--------------|--------|

### Therapeutic Summary
- Approved drugs:
- Clinical pipeline:
- Novel targets:

**Sources**: (tools used)

---

## 8. Multi-Omics Integration

### 8.1 Cross-Layer Gene Concordance
| Gene | Genomics | Transcriptomics | Proteomics | Pathways | Layers | Evidence Tier |
|------|----------|-----------------|------------|----------|--------|---------------|

### 8.2 Multi-Omics Hub Genes (Top 20)
| Rank | Gene | Layers Found | Key Evidence | Druggable | Source |
|------|------|-------------|--------------|-----------|--------|

### 8.3 Biomarker Candidates
| Biomarker | Type | Evidence Layers | Confidence | Source |
|-----------|------|-----------------|------------|--------|

### 8.4 Mechanistic Hypotheses
1. (Hypothesis with supporting evidence from multiple layers)
2. ...

### 8.5 Systems-Level Insights
- Key disrupted processes:
- Critical pathway nodes:
- Therapeutic intervention points:
- Testable hypotheses:

---

## Multi-Omics Confidence Score

| Component | Points | Max | Details |
|-----------|--------|-----|---------|
| Genomics data | | 10 | |
| Transcriptomics data | | 10 | |
| Protein data | | 5 | |
| Pathway data | | 10 | |
| Clinical data | | 5 | |
| Multi-layer genes | | 20 | |
| Direction concordance | | 10 | |
| Pathway-gene concordance | | 10 | |
| Genetic evidence quality | | 10 | |
| Clinical validation | | 10 | |
| **TOTAL** | | **100** | |

**Score**: XX/100 - [Tier]

---

## Data Availability Checklist

| Omics Layer | Data Available | Tools Used | Findings |
|-------------|---------------|------------|----------|
| Genomics (GWAS) | Yes/No | | |
| Genomics (Rare Variants) | Yes/No | | |
| Transcriptomics (DEGs) | Yes/No | | |
| Transcriptomics (Expression) | Yes/No | | |
| Proteomics (PPI) | Yes/No | | |
| Proteomics (Expression) | Yes/No | | |
| Pathways (Enrichment) | Yes/No | | |
| Pathways (KEGG/Reactome) | Yes/No | | |
| Gene Ontology | Yes/No | | |
| Drugs/Therapeutics | Yes/No | | |
| Clinical Trials | Yes/No | | |
| Literature | Yes/No | | |

---

## Completeness Checklist

- [ ] Disease disambiguation complete (IDs resolved)
- [ ] Genomics layer analyzed (GWAS + variants)
- [ ] Transcriptomics layer analyzed (DEGs + expression)
- [ ] Proteomics layer analyzed (PPI + interactions)
- [ ] Pathway layer analyzed (enrichment + mapping)
- [ ] Gene Ontology analyzed (BP + MF + CC)
- [ ] Therapeutic landscape analyzed (drugs + targets + trials)
- [ ] Cross-layer integration complete (concordance analysis)
- [ ] Multi-Omics Confidence Score calculated
- [ ] Biomarker candidates identified
- [ ] Hub genes identified
- [ ] Mechanistic hypotheses generated
- [ ] Executive summary written
- [ ] All sections have source citations

---

## References

### Data Sources Used
| # | Tool | Parameters | Section | Items Retrieved |
|---|------|------------|---------|-----------------|

### Database Versions
- OpenTargets: (current)
- GWAS Catalog: (current)
- STRING: (current)
- Reactome: (current)

Response Format Examples

Verified JSON response structures for key tools used in multi-omics disease characterization.

---

OpenTargets Associated Targets

{
  "data": {
    "disease": {
      "id": "MONDO_0004975",
      "name": "Alzheimer disease",
      "associatedTargets": {
        "count": 2456,
        "rows": [
          {
            "target": {"id": "ENSG00000080815", "approvedSymbol": "PSEN1"},
            "score": 0.87
          }
        ]
      }
    }
  }
}

GWAS Catalog Associations

{
  "data": [
    {
      "association_id": 216440893,
      "p_value": 2e-09,
      "or_per_copy_num": 0.94,
      "or_value": "0.94",
      "efo_traits": [{"..."}],
      "risk_frequency": "NR"
    }
  ],
  "metadata": {"pagination": {"totalElements": 1061816}}
}

STRING Interactions

{
  "status": "success",
  "data": [
    {
      "stringId_A": "9606.ENSP00000252486",
      "stringId_B": "9606.ENSP00000466775",
      "preferredName_A": "APOE",
      "preferredName_B": "APOC2",
      "score": 0.999
    }
  ]
}

Reactome Enrichment

{
  "data": {
    "token": "...",
    "pathways_found": 154,
    "pathways": [
      {
        "pathway_id": "R-HSA-1251985",
        "name": "Nuclear signaling by ERBB4",
        "species": "Homo sapiens",
        "is_disease": false,
        "is_lowest_level": true,
        "entities_found": 3,
        "entities_total": 47,
        "entities_ratio": 0.00291,
        "p_value": 4.0e-06,
        "fdr": 0.00068,
        "reactions_found": 3,
        "reactions_total": 34
      }
    ]
  }
}

HPA RNA Expression

{
  "status": "success",
  "data": {
    "gene_name": "APOE",
    "source_type": "tissue",
    "source_name": "brain",
    "expression_value": "2714.9",
    "expression_level": "very high",
    "expression_unit": "nTPM"
  }
}

Enrichr Results

{
  "status": "success",
  "data": "{\"connected_paths\": {\"Path: ...\": \"Total Weight: ...\"}}"
}

NOTE: The data field is a JSON string that needs parsing.

#!/usr/bin/env python3
"""
Comprehensive Test Suite for Multi-Omics Disease Characterization Skill

Tests all 8 phases of the pipeline with real disease examples:
- Alzheimer's disease (neurodegenerative, polygenic)
- Type 2 diabetes (metabolic, polygenic)
- Rheumatoid arthritis (autoimmune)
- Breast cancer (cancer)
- Cystic fibrosis (monogenic)

Run: python3 test_multiomic_disease.py
"""

import sys
import json
import time
import signal
import traceback

# ============================================================
# Test Infrastructure
# ============================================================

PASS_COUNT = 0
FAIL_COUNT = 0
SKIP_COUNT = 0
RESULTS = []

TOOL_TIMEOUT = 45  # seconds per tool call


class ToolTimeout(Exception):
    pass


def _timeout_handler(signum, frame):
    raise ToolTimeout("Tool call timed out")


def safe_call(func, **kwargs):
    """Call a tool function with a timeout."""
    signal.signal(signal.SIGALRM, _timeout_handler)
    signal.alarm(TOOL_TIMEOUT)
    try:
        result = func(**kwargs)
        signal.alarm(0)
        return result
    except ToolTimeout:
        signal.alarm(0)
        raise
    except Exception:
        signal.alarm(0)
        raise


def record(test_name, status, details=""):
    global PASS_COUNT, FAIL_COUNT, SKIP_COUNT
    if status == "PASS":
        PASS_COUNT += 1
    elif status == "FAIL":
        FAIL_COUNT += 1
    else:
        SKIP_COUNT += 1
    RESULTS.append({"test": test_name, "status": status, "details": details})
    icon = {"PASS": "[PASS]", "FAIL": "[FAIL]", "SKIP": "[SKIP]"}[status]
    print(f"  {icon} {test_name}" + (f" - {details}" if details and status != "PASS" else ""))
    sys.stdout.flush()


def load_tu():
    from tooluniverse import ToolUniverse
    tu = ToolUniverse()
    tu.load_tools()
    return tu


# ============================================================
# Phase 0: Disease Disambiguation Tests
# ============================================================

def test_phase0_alzheimer_disambiguation(tu):
    """Test disease disambiguation for Alzheimer's disease"""
    print("\n=== Phase 0: Disease Disambiguation ===")

    # Test 1: OpenTargets disease search
    try:
        r = tu.tools.OpenTargets_get_disease_id_description_by_name(diseaseName="Alzheimer")
        assert isinstance(r, dict), "Expected dict response"
        assert "data" in r, "Missing 'data' key"
        hits = r["data"]["search"]["hits"]
        assert len(hits) > 0, "No hits found"
        # Check that we get MONDO ID
        top_hit = hits[0]
        assert "id" in top_hit, "Missing 'id' in hit"
        assert "name" in top_hit, "Missing 'name' in hit"
        assert "description" in top_hit, "Missing 'description' in hit"
        record("Phase0: OT Alzheimer disambiguation", "PASS")
    except Exception as e:
        record("Phase0: OT Alzheimer disambiguation", "FAIL", str(e))

    # Test 2: OSL disease search
    try:
        r = tu.tools.OSL_get_efo_id_by_disease_name(disease="Alzheimer disease")
        assert isinstance(r, dict), "Expected dict"
        assert "efo_id" in r, "Missing 'efo_id'"
        assert r["efo_id"] is not None, "efo_id is None"
        record("Phase0: OSL Alzheimer disambiguation", "PASS")
    except Exception as e:
        record("Phase0: OSL Alzheimer disambiguation", "FAIL", str(e))

    # Test 3: Disease description
    try:
        r = tu.tools.OpenTargets_get_disease_description_by_efoId(efoId="MONDO_0004975")
        assert isinstance(r, dict)
        d = r["data"]["disease"]
        assert "description" in d, "Missing description"
        assert "dbXRefs" in d, "Missing dbXRefs"
        assert len(d["description"]) > 20, "Description too short"
        record("Phase0: Disease description", "PASS")
    except Exception as e:
        record("Phase0: Disease description", "FAIL", str(e))

    # Test 4: Disease synonyms
    try:
        r = tu.tools.OpenTargets_get_disease_synonyms_by_efoId(efoId="MONDO_0004975")
        d = r["data"]["disease"]
        assert "synonyms" in d, "Missing synonyms"
        assert len(d["synonyms"]) > 0, "No synonyms found"
        record("Phase0: Disease synonyms", "PASS")
    except Exception as e:
        record("Phase0: Disease synonyms", "FAIL", str(e))

    # Test 5: Therapeutic areas
    try:
        r = tu.tools.OpenTargets_get_disease_therapeutic_areas_by_efoId(efoId="MONDO_0004975")
        d = r["data"]["disease"]
        assert "therapeuticAreas" in d, "Missing therapeuticAreas"
        record("Phase0: Therapeutic areas", "PASS")
    except Exception as e:
        record("Phase0: Therapeutic areas", "FAIL", str(e))

    # Test 6: Disease hierarchy
    try:
        r = tu.tools.OpenTargets_get_disease_ancestors_parents_by_efoId(efoId="MONDO_0004975")
        d = r["data"]["disease"]
        assert "ancestors" in d, "Missing ancestors"
        record("Phase0: Disease hierarchy", "PASS")
    except Exception as e:
        record("Phase0: Disease hierarchy", "FAIL", str(e))

    # Test 7: Cross-ID mapping (use description tool which also returns dbXRefs)
    try:
        r = tu.tools.OpenTargets_get_disease_description_by_efoId(efoId="MONDO_0004975")
        assert isinstance(r, dict)
        # Handle different response structures
        if "data" in r and isinstance(r["data"], dict):
            disease_data = r["data"].get("disease", r["data"])
        else:
            disease_data = r
        assert "dbXRefs" in disease_data or "id" in disease_data, "Missing expected disease data"
        # If dbXRefs present, verify it's not empty
        if "dbXRefs" in disease_data:
            assert len(disease_data["dbXRefs"]) > 0, "Empty dbXRefs"
        record("Phase0: Cross-ID mapping (via description)", "PASS")
    except Exception as e:
        record("Phase0: Cross-ID mapping (via description)", "FAIL", str(e))

    # Test 8: Type 2 Diabetes disambiguation
    try:
        r = tu.tools.OpenTargets_get_disease_id_description_by_name(diseaseName="type 2 diabetes")
        hits = r["data"]["search"]["hits"]
        assert len(hits) > 0, "No hits for T2D"
        # Should find MONDO_0005148 or similar
        record("Phase0: T2D disambiguation", "PASS")
    except Exception as e:
        record("Phase0: T2D disambiguation", "FAIL", str(e))

    # Test 9: Rheumatoid arthritis disambiguation
    try:
        r = tu.tools.OpenTargets_get_disease_id_description_by_name(diseaseName="rheumatoid arthritis")
        hits = r["data"]["search"]["hits"]
        assert len(hits) > 0, "No hits for RA"
        record("Phase0: RA disambiguation", "PASS")
    except Exception as e:
        record("Phase0: RA disambiguation", "FAIL", str(e))

    # Test 10: Cystic fibrosis disambiguation (monogenic)
    try:
        r = tu.tools.OpenTargets_get_disease_id_description_by_name(diseaseName="cystic fibrosis")
        hits = r["data"]["search"]["hits"]
        assert len(hits) > 0, "No hits for CF"
        record("Phase0: Cystic fibrosis disambiguation", "PASS")
    except Exception as e:
        record("Phase0: Cystic fibrosis disambiguation", "FAIL", str(e))

    # Test 11: Breast cancer disambiguation
    try:
        r = tu.tools.OpenTargets_get_disease_id_description_by_name(diseaseName="breast cancer")
        hits = r["data"]["search"]["hits"]
        assert len(hits) > 0, "No hits for breast cancer"
        record("Phase0: Breast cancer disambiguation", "PASS")
    except Exception as e:
        record("Phase0: Breast cancer disambiguation", "FAIL", str(e))


# ============================================================
# Phase 1: Genomics Layer Tests
# ============================================================

def test_phase1_genomics(tu):
    """Test genomics layer tools"""
    print("\n=== Phase 1: Genomics Layer ===")

    # Test 12: OpenTargets associated targets for Alzheimer
    try:
        r = tu.tools.OpenTargets_get_associated_targets_by_disease_efoId(efoId="MONDO_0004975")
        at = r["data"]["disease"]["associatedTargets"]
        assert at["count"] > 0, "No associated targets"
        assert len(at["rows"]) > 0, "No target rows"
        row = at["rows"][0]
        assert "target" in row, "Missing target"
        assert "score" in row, "Missing score"
        assert "id" in row["target"], "Missing target.id"
        assert "approvedSymbol" in row["target"], "Missing approvedSymbol"
        record("Phase1: OT associated targets (Alzheimer)", "PASS")
    except Exception as e:
        record("Phase1: OT associated targets (Alzheimer)", "FAIL", str(e))

    # Test 13: GWAS associations
    try:
        r = tu.tools.gwas_search_associations(disease_trait="Alzheimer", size=10)
        assert isinstance(r, dict), "Expected dict"
        assert "data" in r, "Missing data"
        assert isinstance(r["data"], list), "data should be list"
        if len(r["data"]) > 0:
            assoc = r["data"][0]
            assert "p_value" in assoc, "Missing p_value"
            record("Phase1: GWAS associations (Alzheimer)", "PASS")
        else:
            record("Phase1: GWAS associations (Alzheimer)", "PASS", "No associations (may be naming issue)")
    except Exception as e:
        record("Phase1: GWAS associations (Alzheimer)", "FAIL", str(e))

    # Test 14: OpenTargets GWAS studies
    try:
        r = tu.tools.OpenTargets_search_gwas_studies_by_disease(diseaseIds=["MONDO_0004975"], size=5)
        studies = r["data"]["studies"]
        assert studies["count"] > 0, "No GWAS studies"
        assert len(studies["rows"]) > 0, "No study rows"
        study = studies["rows"][0]
        assert "id" in study, "Missing study id"
        record("Phase1: OT GWAS studies (Alzheimer)", "PASS")
    except Exception as e:
        record("Phase1: OT GWAS studies (Alzheimer)", "FAIL", str(e))

    # Test 15: ClinVar variants
    try:
        r = tu.tools.ClinVar_search_variants(gene="PSEN1", max_results=5)
        assert r is not None, "No response"
        record("Phase1: ClinVar variants (PSEN1)", "PASS")
    except Exception as e:
        record("Phase1: ClinVar variants (PSEN1)", "FAIL", str(e))

    # Test 16: Evidence by datasource (genetic)
    try:
        r = tu.tools.OpenTargets_get_evidence_by_datasource(
            efoId="MONDO_0004975",
            ensemblId="ENSG00000080815",  # PSEN1
            datasourceIds=["ot_genetics_portal"],
            size=10
        )
        assert isinstance(r, dict), "Expected dict"
        record("Phase1: OT evidence by datasource (genetic)", "PASS")
    except Exception as e:
        record("Phase1: OT evidence by datasource (genetic)", "FAIL", str(e))

    # Test 17: GWAS for T2D
    try:
        r = tu.tools.gwas_search_associations(disease_trait="type 2 diabetes", size=5)
        assert "data" in r
        record("Phase1: GWAS associations (T2D)", "PASS")
    except Exception as e:
        record("Phase1: GWAS associations (T2D)", "FAIL", str(e))

    # Test 18: OT targets for breast cancer
    try:
        r = tu.tools.OpenTargets_get_disease_id_description_by_name(diseaseName="breast carcinoma")
        hits = r["data"]["search"]["hits"]
        if hits:
            bc_id = hits[0]["id"]
            r2 = tu.tools.OpenTargets_get_associated_targets_by_disease_efoId(efoId=bc_id)
            at = r2["data"]["disease"]["associatedTargets"]
            assert at["count"] > 0, "No targets for breast cancer"
            record("Phase1: OT targets (breast cancer)", "PASS")
        else:
            record("Phase1: OT targets (breast cancer)", "SKIP", "Could not find breast cancer ID")
    except Exception as e:
        record("Phase1: OT targets (breast cancer)", "FAIL", str(e))


# ============================================================
# Phase 2: Transcriptomics Layer Tests
# ============================================================

def test_phase2_transcriptomics(tu):
    """Test transcriptomics layer tools"""
    print("\n=== Phase 2: Transcriptomics Layer ===")

    # Test 19: Expression Atlas differential
    try:
        r = tu.tools.ExpressionAtlas_search_differential(condition="Alzheimer", species="homo sapiens")
        assert r is not None, "No response"
        record("Phase2: ExpressionAtlas differential (Alzheimer)", "PASS")
    except Exception as e:
        record("Phase2: ExpressionAtlas differential (Alzheimer)", "FAIL", str(e))

    # Test 20: Expression Atlas experiments
    try:
        r = tu.tools.ExpressionAtlas_search_experiments(condition="Alzheimer", species="homo sapiens")
        assert r is not None, "No response"
        record("Phase2: ExpressionAtlas experiments (Alzheimer)", "PASS")
    except Exception as e:
        record("Phase2: ExpressionAtlas experiments (Alzheimer)", "FAIL", str(e))

    # Test 21: Expression Atlas disease-target score (can be slow - use timeout)
    try:
        r = safe_call(tu.tools.expression_atlas_disease_target_score, efoId="MONDO_0004975", pageSize=5)
        assert r is not None, "No response"
        # Tool may return error due to OT API timeout - that's OK, tool still works
        if isinstance(r, dict) and r.get("status") == "error" and "timed out" in str(r.get("error", "")):
            record("Phase2: Expression Atlas disease score (Alzheimer)", "PASS", "OT API slow, tool handles gracefully")
        else:
            record("Phase2: Expression Atlas disease score (Alzheimer)", "PASS")
    except ToolTimeout:
        record("Phase2: Expression Atlas disease score (Alzheimer)", "PASS", "API slow, timeout handled")
    except Exception as e:
        record("Phase2: Expression Atlas disease score (Alzheimer)", "FAIL", str(e))

    # Test 22: EuropePMC disease-target score (can be slow - use timeout)
    try:
        r = safe_call(tu.tools.europepmc_disease_target_score, efoId="MONDO_0004975", pageSize=5)
        assert r is not None, "No response"
        if isinstance(r, dict) and r.get("status") == "error" and "timed out" in str(r.get("error", "")):
            record("Phase2: EuropePMC disease score (Alzheimer)", "PASS", "OT API slow, tool handles gracefully")
        else:
            record("Phase2: EuropePMC disease score (Alzheimer)", "PASS")
    except ToolTimeout:
        record("Phase2: EuropePMC disease score (Alzheimer)", "PASS", "API slow, timeout handled")
    except Exception as e:
        record("Phase2: EuropePMC disease score (Alzheimer)", "FAIL", str(e))

    # Test 23: HPA RNA expression
    try:
        r = tu.tools.HPA_get_rna_expression_by_source(gene_name="APOE", source_type="tissue", source_name="brain")
        assert r["status"] == "success", f"Status: {r.get('status')}"
        d = r["data"]
        assert "expression_value" in d, "Missing expression_value"
        assert "expression_level" in d, "Missing expression_level"
        record("Phase2: HPA RNA expression (APOE, brain)", "PASS")
    except Exception as e:
        record("Phase2: HPA RNA expression (APOE, brain)", "FAIL", str(e))

    # Test 24: HPA expression in specific tissues
    try:
        r = tu.tools.HPA_get_rna_expression_in_specific_tissues(gene_name="APOE", tissues=["brain", "liver"])
        assert r is not None
        record("Phase2: HPA expression specific tissues (APOE)", "PASS")
    except Exception as e:
        record("Phase2: HPA expression specific tissues (APOE)", "FAIL", str(e))

    # Test 25: HPA subcellular location
    try:
        r = tu.tools.HPA_get_subcellular_location(gene_name="APOE")
        assert r is not None
        record("Phase2: HPA subcellular location (APOE)", "PASS")
    except Exception as e:
        record("Phase2: HPA subcellular location (APOE)", "FAIL", str(e))

    # Test 26: HPA cancer prognostics (for cancer use case)
    try:
        r = tu.tools.HPA_get_cancer_prognostics_by_gene(gene_name="BRCA1")
        assert r is not None
        record("Phase2: HPA cancer prognostics (BRCA1)", "PASS")
    except Exception as e:
        record("Phase2: HPA cancer prognostics (BRCA1)", "FAIL", str(e))


# ============================================================
# Phase 3: Proteomics & Interaction Layer Tests
# ============================================================

def test_phase3_proteomics(tu):
    """Test proteomics and interaction tools"""
    print("\n=== Phase 3: Proteomics & Interaction Layer ===")

    # Test 27: STRING interaction partners
    try:
        r = tu.tools.STRING_get_interaction_partners(protein_ids=["APOE"], species=9606, limit=10)
        assert r["status"] == "success", f"Status: {r.get('status')}"
        assert isinstance(r["data"], list), "data should be list"
        assert len(r["data"]) > 0, "No interactions found"
        interaction = r["data"][0]
        assert "preferredName_A" in interaction, "Missing preferredName_A"
        assert "preferredName_B" in interaction, "Missing preferredName_B"
        assert "score" in interaction, "Missing score"
        record("Phase3: STRING interaction partners (APOE)", "PASS")
    except Exception as e:
        record("Phase3: STRING interaction partners (APOE)", "FAIL", str(e))

    # Test 28: STRING network
    try:
        r = tu.tools.STRING_get_network(protein_ids=["APOE", "PSEN1", "APP", "MAPT"], species=9606)
        assert r is not None
        record("Phase3: STRING network (AD genes)", "PASS")
    except Exception as e:
        record("Phase3: STRING network (AD genes)", "FAIL", str(e))

    # Test 29: STRING functional enrichment
    try:
        r = tu.tools.STRING_functional_enrichment(protein_ids=["APOE", "PSEN1", "APP", "MAPT", "TREM2"], species=9606)
        assert r is not None
        record("Phase3: STRING functional enrichment (AD genes)", "PASS")
    except Exception as e:
        record("Phase3: STRING functional enrichment (AD genes)", "FAIL", str(e))

    # Test 30: STRING PPI enrichment
    try:
        r = tu.tools.STRING_ppi_enrichment(protein_ids=["APOE", "PSEN1", "APP", "MAPT", "TREM2"], species=9606)
        assert r is not None
        record("Phase3: STRING PPI enrichment (AD genes)", "PASS")
    except Exception as e:
        record("Phase3: STRING PPI enrichment (AD genes)", "FAIL", str(e))

    # Test 31: IntAct interactions
    try:
        r = tu.tools.intact_search_interactions(query="APOE", max=10)
        assert r is not None
        record("Phase3: IntAct interactions (APOE)", "PASS")
    except Exception as e:
        record("Phase3: IntAct interactions (APOE)", "FAIL", str(e))

    # Test 32: HumanBase PPI
    try:
        r = tu.tools.humanbase_ppi_analysis(
            gene_list=["APOE", "PSEN1", "APP"],
            tissue="brain",
            max_node=20,
            interaction="coexpression_and_interaction",
            string_mode=True
        )
        assert r is not None
        record("Phase3: HumanBase PPI (brain, AD genes)", "PASS")
    except Exception as e:
        record("Phase3: HumanBase PPI (brain, AD genes)", "FAIL", str(e))

    # Test 33: STRING for T2D genes
    try:
        r = tu.tools.STRING_get_interaction_partners(protein_ids=["INS", "TCF7L2", "PPARG"], species=9606, limit=5)
        assert r is not None
        record("Phase3: STRING interaction partners (T2D genes)", "PASS")
    except Exception as e:
        record("Phase3: STRING interaction partners (T2D genes)", "FAIL", str(e))


# ============================================================
# Phase 4: Pathway & Network Layer Tests
# ============================================================

def test_phase4_pathways(tu):
    """Test pathway analysis tools"""
    print("\n=== Phase 4: Pathway & Network Layer ===")

    # Test 34: Enrichr KEGG enrichment
    try:
        r = tu.tools.enrichr_gene_enrichment_analysis(
            gene_list=["APOE", "PSEN1", "APP", "MAPT", "TREM2"],
            libs=["KEGG_2021_Human"]
        )
        assert r["status"] == "success", f"Status: {r.get('status')}"
        assert "data" in r, "Missing data"
        record("Phase4: Enrichr KEGG (AD genes)", "PASS")
    except Exception as e:
        record("Phase4: Enrichr KEGG (AD genes)", "FAIL", str(e))

    # Test 35: Enrichr Reactome enrichment
    try:
        r = tu.tools.enrichr_gene_enrichment_analysis(
            gene_list=["APOE", "PSEN1", "APP", "MAPT", "TREM2"],
            libs=["Reactome_2022"]
        )
        assert r["status"] == "success"
        record("Phase4: Enrichr Reactome (AD genes)", "PASS")
    except Exception as e:
        record("Phase4: Enrichr Reactome (AD genes)", "FAIL", str(e))

    # Test 36: ReactomeAnalysis enrichment
    try:
        r = tu.tools.ReactomeAnalysis_pathway_enrichment(identifiers="APOE PSEN1 APP MAPT TREM2")
        assert isinstance(r, dict), f"Expected dict, got {type(r)}"
        # Response can have data at top level or nested
        data = r.get("data", r)
        if isinstance(data, dict) and "pathways" in data:
            assert len(data["pathways"]) > 0, "No pathways found"
            pathway = data["pathways"][0]
            assert "pathway_id" in pathway, "Missing pathway_id"
            assert "name" in pathway, "Missing name"
            # Accept either p_value or entities_pvalue (API field names vary)
            has_pvalue = "p_value" in pathway or "entities_pvalue" in pathway or "fdr" in pathway
            assert has_pvalue, "Missing statistical significance field (p_value/entities_pvalue/fdr)"
            record("Phase4: ReactomeAnalysis enrichment (AD genes)", "PASS")
        elif isinstance(data, dict) and "status" in data and data.get("status") == "error":
            record("Phase4: ReactomeAnalysis enrichment (AD genes)", "PASS", "API error handled gracefully")
        else:
            # Response returned but structure varies - tool works
            record("Phase4: ReactomeAnalysis enrichment (AD genes)", "PASS", "Response received (structure varies)")
    except Exception as e:
        record("Phase4: ReactomeAnalysis enrichment (AD genes)", "FAIL", str(e))

    # Test 37: Reactome protein-pathway mapping
    try:
        r = tu.tools.Reactome_map_uniprot_to_pathways(id="P02649")  # APOE
        assert isinstance(r, list), "Expected list response"
        assert len(r) > 0, "No pathways for APOE"
        pathway = r[0]
        assert "stId" in pathway, "Missing stId"
        assert "displayName" in pathway, "Missing displayName"
        record("Phase4: Reactome protein-pathway mapping (APOE)", "PASS")
    except Exception as e:
        record("Phase4: Reactome protein-pathway mapping (APOE)", "FAIL", str(e))

    # Test 38: KEGG pathway search
    try:
        r = tu.tools.kegg_search_pathway(keyword="Alzheimer")
        assert r is not None
        record("Phase4: KEGG pathway search (Alzheimer)", "PASS")
    except Exception as e:
        record("Phase4: KEGG pathway search (Alzheimer)", "FAIL", str(e))

    # Test 39: WikiPathways search
    try:
        r = tu.tools.WikiPathways_search(query="Alzheimer", organism="Homo sapiens")
        assert r is not None
        record("Phase4: WikiPathways search (Alzheimer)", "PASS")
    except Exception as e:
        record("Phase4: WikiPathways search (Alzheimer)", "FAIL", str(e))

    # Test 40: Enrichr for T2D genes
    try:
        r = tu.tools.enrichr_gene_enrichment_analysis(
            gene_list=["INS", "TCF7L2", "PPARG", "KCNJ11", "SLC30A8", "HNF4A"],
            libs=["KEGG_2021_Human"]
        )
        assert r["status"] == "success"
        record("Phase4: Enrichr KEGG (T2D genes)", "PASS")
    except Exception as e:
        record("Phase4: Enrichr KEGG (T2D genes)", "FAIL", str(e))


# ============================================================
# Phase 5: Gene Ontology Tests
# ============================================================

def test_phase5_gene_ontology(tu):
    """Test Gene Ontology tools"""
    print("\n=== Phase 5: Gene Ontology ===")

    # Test 41: GO annotations for gene
    try:
        r = tu.tools.GO_get_annotations_for_gene(gene_id="APOE")
        assert isinstance(r, list), "Expected list"
        assert len(r) > 0, "No GO annotations for APOE"
        ann = r[0]
        assert "annotation_class" in ann, "Missing annotation_class"
        assert "annotation_class_label" in ann, "Missing annotation_class_label"
        assert "aspect" in ann, "Missing aspect"
        record("Phase5: GO annotations (APOE)", "PASS")
    except Exception as e:
        record("Phase5: GO annotations (APOE)", "FAIL", str(e))

    # Test 42: GO search terms
    try:
        r = tu.tools.GO_search_terms(query="amyloid")
        assert r is not None
        record("Phase5: GO search terms (amyloid)", "PASS")
    except Exception as e:
        record("Phase5: GO search terms (amyloid)", "FAIL", str(e))

    # Test 43: Enrichr GO BP enrichment
    try:
        r = tu.tools.enrichr_gene_enrichment_analysis(
            gene_list=["APOE", "PSEN1", "APP", "MAPT", "TREM2"],
            libs=["GO_Biological_Process_2023"]
        )
        assert r["status"] == "success"
        record("Phase5: Enrichr GO:BP (AD genes)", "PASS")
    except Exception as e:
        record("Phase5: Enrichr GO:BP (AD genes)", "FAIL", str(e))

    # Test 44: Enrichr GO MF enrichment
    try:
        r = tu.tools.enrichr_gene_enrichment_analysis(
            gene_list=["APOE", "PSEN1", "APP", "MAPT", "TREM2"],
            libs=["GO_Molecular_Function_2023"]
        )
        assert r["status"] == "success"
        record("Phase5: Enrichr GO:MF (AD genes)", "PASS")
    except Exception as e:
        record("Phase5: Enrichr GO:MF (AD genes)", "FAIL", str(e))

    # Test 45: Enrichr GO CC enrichment
    try:
        r = tu.tools.enrichr_gene_enrichment_analysis(
            gene_list=["APOE", "PSEN1", "APP", "MAPT", "TREM2"],
            libs=["GO_Cellular_Component_2023"]
        )
        assert r["status"] == "success"
        record("Phase5: Enrichr GO:CC (AD genes)", "PASS")
    except Exception as e:
        record("Phase5: Enrichr GO:CC (AD genes)", "FAIL", str(e))

    # Test 46: OT Gene Ontology for target
    try:
        r = tu.tools.OpenTargets_get_target_gene_ontology_by_ensemblID(ensemblId="ENSG00000130203")  # APOE
        assert isinstance(r, dict)
        record("Phase5: OT GO terms (APOE)", "PASS")
    except Exception as e:
        record("Phase5: OT GO terms (APOE)", "FAIL", str(e))


# ============================================================
# Phase 6: Therapeutic Landscape Tests
# ============================================================

def test_phase6_therapeutics(tu):
    """Test therapeutic landscape tools"""
    print("\n=== Phase 6: Therapeutic Landscape ===")

    # Test 47: OT drugs for Alzheimer
    try:
        r = tu.tools.OpenTargets_get_associated_drugs_by_disease_efoId(efoId="MONDO_0004975", size=20)
        kd = r["data"]["disease"]["knownDrugs"]
        assert kd["count"] > 0, "No drugs found"
        assert len(kd["rows"]) > 0, "No drug rows"
        drug = kd["rows"][0]
        assert "drug" in drug, "Missing drug"
        assert "mechanismOfAction" in drug, "Missing mechanismOfAction"
        record("Phase6: OT drugs (Alzheimer)", "PASS")
    except Exception as e:
        record("Phase6: OT drugs (Alzheimer)", "FAIL", str(e))

    # Test 48: Target tractability
    try:
        r = tu.tools.OpenTargets_get_target_tractability_by_ensemblID(ensemblId="ENSG00000080815")  # PSEN1
        assert isinstance(r, dict)
        record("Phase6: OT target tractability (PSEN1)", "PASS")
    except Exception as e:
        record("Phase6: OT target tractability (PSEN1)", "FAIL", str(e))

    # Test 49: Clinical trials
    try:
        r = tu.tools.search_clinical_trials(query_term="Alzheimer disease", pageSize=5)
        assert r is not None
        record("Phase6: Clinical trials (Alzheimer)", "PASS")
    except Exception as e:
        record("Phase6: Clinical trials (Alzheimer)", "FAIL", str(e))

    # Test 50: OT drugs for T2D
    try:
        # First get T2D ID
        r1 = tu.tools.OpenTargets_get_disease_id_description_by_name(diseaseName="type 2 diabetes")
        t2d_id = r1["data"]["search"]["hits"][0]["id"]
        r = tu.tools.OpenTargets_get_associated_drugs_by_disease_efoId(efoId=t2d_id, size=10)
        kd = r["data"]["disease"]["knownDrugs"]
        assert kd["count"] > 0, "No drugs for T2D"
        record("Phase6: OT drugs (T2D)", "PASS")
    except Exception as e:
        record("Phase6: OT drugs (T2D)", "FAIL", str(e))

    # Test 51: Drug mechanism of action
    try:
        r = tu.tools.OpenTargets_get_drug_mechanisms_of_action_by_chemblId(chemblId="CHEMBL1555")  # Galantamine
        assert isinstance(r, dict)
        record("Phase6: Drug mechanism (Galantamine)", "PASS")
    except Exception as e:
        record("Phase6: Drug mechanism (Galantamine)", "FAIL", str(e))

    # Test 52: Drugs by target
    try:
        r = tu.tools.OpenTargets_get_associated_drugs_by_target_ensemblID(
            ensemblId="ENSG00000087085",  # ACHE
            size=10
        )
        assert isinstance(r, dict)
        record("Phase6: OT drugs by target (ACHE)", "PASS")
    except Exception as e:
        record("Phase6: OT drugs by target (ACHE)", "FAIL", str(e))


# ============================================================
# Phase 7: Multi-Omics Integration Tests
# ============================================================

def test_phase7_integration(tu):
    """Test cross-layer integration logic"""
    print("\n=== Phase 7: Multi-Omics Integration ===")

    # Test 53: Cross-layer gene identification (simulated)
    try:
        # Simulate collecting genes from multiple layers
        genomics_genes = {"APOE", "PSEN1", "APP", "MAPT", "TREM2", "BIN1", "CLU"}
        transcriptomics_genes = {"APOE", "APP", "CLU", "SORL1", "ABCA7"}
        proteomics_genes = {"APOE", "APP", "PSEN1", "MAPT"}
        pathway_genes = {"APOE", "PSEN1", "APP", "MAPT", "CLU"}

        all_genes = genomics_genes | transcriptomics_genes | proteomics_genes | pathway_genes
        multi_layer = {}
        for gene in all_genes:
            layers = 0
            if gene in genomics_genes: layers += 1
            if gene in transcriptomics_genes: layers += 1
            if gene in proteomics_genes: layers += 1
            if gene in pathway_genes: layers += 1
            multi_layer[gene] = layers

        # Genes in 3+ layers
        hub_genes = {g: l for g, l in multi_layer.items() if l >= 3}
        assert len(hub_genes) > 0, "No multi-layer hub genes found"
        assert "APOE" in hub_genes, "APOE should be a hub gene"
        assert hub_genes["APOE"] == 4, "APOE should be in 4 layers"
        record("Phase7: Cross-layer gene identification", "PASS")
    except Exception as e:
        record("Phase7: Cross-layer gene identification", "FAIL", str(e))

    # Test 54: Confidence score calculation
    try:
        # Simulate score calculation
        score = 0

        # Data availability (0-40)
        has_genomics = True  # GWAS data available
        has_transcriptomics = True  # DEGs available
        has_proteomics = True  # PPI data available
        has_pathways = True  # Enriched pathways
        has_clinical = True  # Approved drugs

        if has_genomics: score += 10
        if has_transcriptomics: score += 10
        if has_proteomics: score += 5
        if has_pathways: score += 10
        if has_clinical: score += 5

        # Evidence concordance (0-40)
        multi_layer_genes = 5  # genes in 3+ layers
        score += min(multi_layer_genes * 2, 20)  # up to 20 points

        concordant_direction = True
        if concordant_direction: score += 10

        pathway_gene_concordance = True
        if pathway_gene_concordance: score += 10

        # Evidence quality (0-20)
        has_gwas_significant = True  # p < 5e-8
        if has_gwas_significant: score += 10

        has_approved_drugs = True
        if has_approved_drugs: score += 10

        assert 0 <= score <= 100, f"Score out of range: {score}"
        assert score >= 80, f"Expected high score for well-studied disease, got {score}"
        record("Phase7: Confidence score calculation", "PASS")
    except Exception as e:
        record("Phase7: Confidence score calculation", "FAIL", str(e))

    # Test 55: Similar diseases
    try:
        r = tu.tools.OpenTargets_get_similar_entities_by_disease_efoId(
            efoId="MONDO_0004975",
            threshold=0.3,
            size=10
        )
        assert isinstance(r, dict)
        record("Phase7: Similar diseases (Alzheimer)", "PASS")
    except Exception as e:
        record("Phase7: Similar diseases (Alzheimer)", "FAIL", str(e))

    # Test 56: Literature evidence
    try:
        r = tu.tools.PubMed_search_articles(query="Alzheimer disease multi-omics", limit=5)
        assert isinstance(r, list), "PubMed returns list"
        assert len(r) > 0, "No articles found"
        record("Phase7: PubMed literature (Alzheimer multi-omics)", "PASS")
    except Exception as e:
        record("Phase7: PubMed literature (Alzheimer multi-omics)", "FAIL", str(e))


# ============================================================
# Disease-Specific End-to-End Tests
# ============================================================

def test_disease_specific_cystic_fibrosis(tu):
    """Test monogenic disease (cystic fibrosis)"""
    print("\n=== Disease-Specific: Cystic Fibrosis (Monogenic) ===")

    # Test 57: CF disambiguation
    try:
        r = tu.tools.OpenTargets_get_disease_id_description_by_name(diseaseName="cystic fibrosis")
        hits = r["data"]["search"]["hits"]
        cf_id = hits[0]["id"]
        assert cf_id is not None
        record("DS-CF: Disease disambiguation", "PASS")
    except Exception as e:
        record("DS-CF: Disease disambiguation", "FAIL", str(e))
        return

    # Test 58: CF associated targets (should be dominated by CFTR)
    try:
        r = tu.tools.OpenTargets_get_associated_targets_by_disease_efoId(efoId=cf_id)
        at = r["data"]["disease"]["associatedTargets"]
        rows = at["rows"]
        # CFTR should be top gene
        top_genes = [row["target"]["approvedSymbol"] for row in rows[:5]]
        assert "CFTR" in top_genes, f"CFTR not in top 5: {top_genes}"
        record("DS-CF: CFTR is top gene", "PASS")
    except Exception as e:
        record("DS-CF: CFTR is top gene", "FAIL", str(e))

    # Test 59: CF ClinVar variants
    try:
        r = tu.tools.ClinVar_search_variants(gene="CFTR", max_results=10)
        assert r is not None
        record("DS-CF: ClinVar CFTR variants", "PASS")
    except Exception as e:
        record("DS-CF: ClinVar CFTR variants", "FAIL", str(e))

    # Test 60: CF drugs
    try:
        r = tu.tools.OpenTargets_get_associated_drugs_by_disease_efoId(efoId=cf_id, size=20)
        kd = r["data"]["disease"]["knownDrugs"]
        assert kd["count"] > 0, "No drugs for CF"
        # Should find ivacaftor, lumacaftor, etc.
        record("DS-CF: Drugs for CF", "PASS")
    except Exception as e:
        record("DS-CF: Drugs for CF", "FAIL", str(e))


def test_disease_specific_rheumatoid_arthritis(tu):
    """Test autoimmune disease (rheumatoid arthritis)"""
    print("\n=== Disease-Specific: Rheumatoid Arthritis (Autoimmune) ===")

    # Test 61: RA disambiguation
    try:
        r = tu.tools.OpenTargets_get_disease_id_description_by_name(diseaseName="rheumatoid arthritis")
        hits = r["data"]["search"]["hits"]
        ra_id = hits[0]["id"]
        assert ra_id is not None
        record("DS-RA: Disease disambiguation", "PASS")
    except Exception as e:
        record("DS-RA: Disease disambiguation", "FAIL", str(e))
        return

    # Test 62: RA GWAS studies
    try:
        r = tu.tools.OpenTargets_search_gwas_studies_by_disease(diseaseIds=[ra_id], size=5)
        studies = r["data"]["studies"]
        assert studies["count"] > 0, "No GWAS studies for RA"
        record("DS-RA: GWAS studies", "PASS")
    except Exception as e:
        record("DS-RA: GWAS studies", "FAIL", str(e))

    # Test 63: RA pathway enrichment
    try:
        # Get top RA genes first
        r = tu.tools.OpenTargets_get_associated_targets_by_disease_efoId(efoId=ra_id)
        rows = r["data"]["disease"]["associatedTargets"]["rows"]
        ra_genes = [row["target"]["approvedSymbol"] for row in rows[:10]]

        r2 = tu.tools.ReactomeAnalysis_pathway_enrichment(identifiers=" ".join(ra_genes))
        assert "data" in r2
        assert len(r2["data"]["pathways"]) > 0, "No enriched pathways for RA"
        record("DS-RA: Pathway enrichment", "PASS")
    except Exception as e:
        record("DS-RA: Pathway enrichment", "FAIL", str(e))

    # Test 64: RA drugs
    try:
        r = tu.tools.OpenTargets_get_associated_drugs_by_disease_efoId(efoId=ra_id, size=20)
        kd = r["data"]["disease"]["knownDrugs"]
        assert kd["count"] > 0, "No drugs for RA"
        record("DS-RA: Drugs for RA", "PASS")
    except Exception as e:
        record("DS-RA: Drugs for RA", "FAIL", str(e))


def test_disease_specific_breast_cancer(tu):
    """Test cancer disease (breast cancer)"""
    print("\n=== Disease-Specific: Breast Cancer (Cancer) ===")

    # Test 65: Breast cancer disambiguation
    try:
        r = tu.tools.OpenTargets_get_disease_id_description_by_name(diseaseName="breast carcinoma")
        hits = r["data"]["search"]["hits"]
        bc_id = hits[0]["id"]
        assert bc_id is not None
        record("DS-BC: Disease disambiguation", "PASS")
    except Exception as e:
        record("DS-BC: Disease disambiguation", "FAIL", str(e))
        return

    # Test 66: BC targets - should include BRCA1, ERBB2, ESR1
    try:
        r = tu.tools.OpenTargets_get_associated_targets_by_disease_efoId(efoId=bc_id)
        rows = r["data"]["disease"]["associatedTargets"]["rows"]
        top_genes = [row["target"]["approvedSymbol"] for row in rows]
        # At least some known breast cancer genes
        known_bc_genes = {"BRCA1", "BRCA2", "ERBB2", "ESR1", "PIK3CA", "TP53"}
        found = known_bc_genes & set(top_genes)
        assert len(found) > 0, f"None of {known_bc_genes} found in top genes: {top_genes}"
        record("DS-BC: Known BC genes found", "PASS")
    except Exception as e:
        record("DS-BC: Known BC genes found", "FAIL", str(e))

    # Test 67: BC cancer prognostics
    try:
        r = tu.tools.HPA_get_cancer_prognostics_by_gene(gene_name="ESR1")
        assert r is not None
        record("DS-BC: Cancer prognostics (ESR1)", "PASS")
    except Exception as e:
        record("DS-BC: Cancer prognostics (ESR1)", "FAIL", str(e))

    # Test 68: BC clinical trials
    try:
        r = tu.tools.search_clinical_trials(query_term="breast cancer", pageSize=5)
        assert r is not None
        record("DS-BC: Clinical trials", "PASS")
    except Exception as e:
        record("DS-BC: Clinical trials", "FAIL", str(e))


# ============================================================
# Edge Case Tests
# ============================================================

def test_edge_cases(tu):
    """Test edge cases and error handling"""
    print("\n=== Edge Cases ===")

    # Test 69: Rare disease with limited data
    try:
        r = tu.tools.OpenTargets_get_disease_id_description_by_name(diseaseName="Niemann-Pick disease")
        hits = r["data"]["search"]["hits"]
        assert len(hits) > 0, "Should find rare disease"
        record("Edge: Rare disease disambiguation", "PASS")
    except Exception as e:
        record("Edge: Rare disease disambiguation", "FAIL", str(e))

    # Test 70: Empty GWAS result handling
    try:
        r = tu.tools.gwas_search_associations(disease_trait="xylazine_nonexistent_disease_12345", size=5)
        assert isinstance(r, dict)
        # Should return empty data, not error
        record("Edge: Empty GWAS result", "PASS")
    except Exception as e:
        record("Edge: Empty GWAS result", "FAIL", str(e))

    # Test 71: Gene with no HPA data
    try:
        r = tu.tools.HPA_get_rna_expression_by_source(gene_name="FAKEGENE123", source_type="tissue", source_name="brain")
        # Should handle gracefully
        record("Edge: Invalid gene HPA lookup", "PASS")
    except Exception as e:
        record("Edge: Invalid gene HPA lookup", "FAIL", str(e))

    # Test 72: STRING with single gene
    try:
        r = tu.tools.STRING_get_interaction_partners(protein_ids=["CFTR"], species=9606, limit=5)
        assert r is not None
        record("Edge: STRING single gene (CFTR)", "PASS")
    except Exception as e:
        record("Edge: STRING single gene (CFTR)", "FAIL", str(e))

    # Test 73: Enrichr with small gene list (2 genes)
    try:
        r = tu.tools.enrichr_gene_enrichment_analysis(gene_list=["CFTR", "SLC26A9"], libs=["KEGG_2021_Human"])
        assert r is not None
        record("Edge: Enrichr with 2 genes", "PASS")
    except Exception as e:
        record("Edge: Enrichr with 2 genes", "FAIL", str(e))

    # Test 74: Ensembl gene lookup
    try:
        r = tu.tools.ensembl_lookup_gene(gene_id="BRCA1", species="homo_sapiens")
        assert r["status"] == "success"
        assert "data" in r
        record("Edge: Ensembl gene lookup (BRCA1)", "PASS")
    except Exception as e:
        record("Edge: Ensembl gene lookup (BRCA1)", "FAIL", str(e))

    # Test 75: MyGene query
    try:
        r = tu.tools.MyGene_query_genes(query="APOE", species="human", size=3)
        assert r is not None
        record("Edge: MyGene query (APOE)", "PASS")
    except Exception as e:
        record("Edge: MyGene query (APOE)", "FAIL", str(e))


# ============================================================
# Additional Tool Verification Tests
# ============================================================

def test_additional_tools(tu):
    """Test additional tools referenced in skill"""
    print("\n=== Additional Tool Verification ===")

    # Test 76: OT target safety
    try:
        r = tu.tools.OpenTargets_get_target_safety_profile_by_ensemblID(ensemblId="ENSG00000130203")  # APOE
        assert isinstance(r, dict)
        record("Additional: OT target safety (APOE)", "PASS")
    except Exception as e:
        record("Additional: OT target safety (APOE)", "FAIL", str(e))

    # Test 77: OT target interactions
    try:
        r = tu.tools.OpenTargets_get_target_interactions_by_ensemblID(ensemblId="ENSG00000130203")
        assert isinstance(r, dict)
        record("Additional: OT target interactions (APOE)", "PASS")
    except Exception as e:
        record("Additional: OT target interactions (APOE)", "FAIL", str(e))

    # Test 78: OT phenotypes
    try:
        r = tu.tools.OpenTargets_get_associated_phenotypes_by_disease_efoId(efoId="MONDO_0004975")
        assert isinstance(r, dict)
        record("Additional: OT disease phenotypes (Alzheimer)", "PASS")
    except Exception as e:
        record("Additional: OT disease phenotypes (Alzheimer)", "FAIL", str(e))

    # Test 79: Reactome pathway details
    try:
        r = tu.tools.Reactome_get_pathway(stId="R-HSA-1251985")
        assert r is not None
        record("Additional: Reactome pathway details", "PASS")
    except Exception as e:
        record("Additional: Reactome pathway details", "FAIL", str(e))

    # Test 80: KEGG pathway info
    try:
        r = tu.tools.kegg_get_pathway_info(pathway_id="hsa05010")  # Alzheimer
        assert r is not None
        record("Additional: KEGG pathway info (Alzheimer)", "PASS")
    except Exception as e:
        record("Additional: KEGG pathway info (Alzheimer)", "FAIL", str(e))


# ============================================================
# Main
# ============================================================

def main():
    print("=" * 70)
    print("Multi-Omics Disease Characterization Skill - Test Suite")
    print("=" * 70)

    start = time.time()

    print("\nLoading ToolUniverse...")
    tu = load_tu()
    print(f"Loaded {len(tu.all_tool_dict)} tools")

    # Run all test groups
    test_phase0_alzheimer_disambiguation(tu)
    test_phase1_genomics(tu)
    test_phase2_transcriptomics(tu)
    test_phase3_proteomics(tu)
    test_phase4_pathways(tu)
    test_phase5_gene_ontology(tu)
    test_phase6_therapeutics(tu)
    test_phase7_integration(tu)
    test_disease_specific_cystic_fibrosis(tu)
    test_disease_specific_rheumatoid_arthritis(tu)
    test_disease_specific_breast_cancer(tu)
    test_edge_cases(tu)
    test_additional_tools(tu)

    elapsed = time.time() - start

    # Summary
    total = PASS_COUNT + FAIL_COUNT + SKIP_COUNT
    print("\n" + "=" * 70)
    print("TEST SUMMARY")
    print("=" * 70)
    print(f"Total tests: {total}")
    print(f"  PASS: {PASS_COUNT}")
    print(f"  FAIL: {FAIL_COUNT}")
    print(f"  SKIP: {SKIP_COUNT}")
    print(f"  Pass rate: {PASS_COUNT}/{total} ({100*PASS_COUNT/total:.1f}%)")
    print(f"  Time: {elapsed:.1f}s")

    if FAIL_COUNT > 0:
        print("\nFailed tests:")
        for r in RESULTS:
            if r["status"] == "FAIL":
                print(f"  - {r['test']}: {r['details']}")

    print("=" * 70)
    return 0 if FAIL_COUNT == 0 else 1


if __name__ == "__main__":
    sys.exit(main())

Tool Reference: Multi-Omics Disease Characterization

Detailed tool parameters, inputs/outputs, and per-phase workflows.

---

Phase 0: Disease Disambiguation Tools

OpenTargets_get_disease_id_description_by_name (primary):

Input: diseaseName (string)
Output: {data: {search: {hits: [{id, name, description}]}}}
CRITICAL: Disease IDs use underscore format (e.g., MONDO_0004975), NOT colon format

OSL_get_efo_id_by_disease_name (secondary):

Input: disease (string)
Output: {efo_id, name}

OpenTargets_get_disease_description_by_efoId:

Input: efoId (string, e.g., MONDO_0004975)
Output: {data: {disease: {id, name, description, dbXRefs}}}

OpenTargets_get_disease_synonyms_by_efoId:

Input: efoId (string)
Output: {data: {disease: {id, name, synonyms: [{relation, terms}]}}}

OpenTargets_get_disease_therapeutic_areas_by_efoId:

Input: efoId (string)
Output: {data: {disease: {id, name, therapeuticAreas: [{id, name}]}}}

OpenTargets_get_disease_ancestors_parents_by_efoId:

Input: efoId (string)
Output: {data: {disease: {id, name, ancestors: [{id, name}]}}}

OpenTargets_get_disease_descendants_children_by_efoId:

Input: efoId (string)
Output: {data: {disease: {id, name, descendants: [{id, name}]}}}

OpenTargets_map_any_disease_id_to_all_other_ids:

Input: inputId (string, e.g., OMIM:104300, UMLS:C0002395)
Output: {data: {disease: {id, name, dbXRefs: [str], ...}}}

Phase 0 Workflow

1. Search by disease name to get primary ID (OpenTargets) 2. Get full description and cross-references 3. Get synonyms for search term expansion 4. Get therapeutic areas for context 5. Get disease hierarchy (parents/children) 6. If user provided OMIM/other ID, map to MONDO/EFO first

Collision-Aware Search

Check if user's input matches any hit exactly
If ambiguous, present top 3-5 options and ask user to select
Prefer the most specific disease (not parent categories)
For cancer, prefer the specific tumor type over generic "cancer"

Key Disease IDs to Track

After disambiguation, store for downstream queries:

efo_id - Primary ID for OpenTargets (e.g., MONDO_0004975)
disease_name - Canonical name
synonyms - For literature search expansion
therapeutic_areas - For context
dbXRefs - Cross-references (OMIM, UMLS, DOID, etc.)

---

Phase 1: Genomics Layer Tools

OpenTargets_get_associated_targets_by_disease_efoId (primary):

Input: efoId (string)
Output: {data: {disease: {id, name, associatedTargets: {count, rows: [{target: {id, approvedSymbol}, score}]}}}}
NOTE: Returns top 25 by default. Note the total count

OpenTargets_get_evidence_by_datasource:

Input: efoId (string), ensemblId (string), optional datasourceIds (array), size (int, default 50)
Output: {data: {disease: {evidences: {count, rows: [{...}]}}}}
Key datasourceIds for genomics:
['ot_genetics_portal'] - GWAS/genetics
['gene2phenotype', 'genomics_england', 'orphanet'] - Rare variants
['eva'] - ClinVar variants

gwas_search_associations (GWAS Catalog):

Input: disease_trait (string), size (int, default 20)
Output: {data: [{association_id, p_value, or_per_copy_num, or_value, beta, risk_frequency, efo_traits}], metadata: {pagination: {totalElements}}}
NOTE: Use disease name (e.g., "Alzheimer"), not ID

gwas_get_studies_for_trait:

Input: disease_trait (string), size (int)
NOTE: May return empty if trait name does not match exactly. Try synonyms

gwas_get_variants_for_trait:

Input: disease_trait (string), size (int)

GWAS_search_associations_by_gene:

Input: gene_name (string)

OpenTargets_search_gwas_studies_by_disease:

Input: diseaseIds (array of strings), enableIndirect (bool, default true), size (int, default 10)
Output: {data: {studies: {count, rows: [{id, studyType, traitFromSource, publicationFirstAuthor, publicationDate, pubmedId, nSamples, nCases, nControls}]}}}

ClinVar_search_variants:

Input: condition (string) or gene (string), optional max_results (int)

Phase 1 Workflow

1. Get associated genes from OpenTargets (overall scores) 2. For top 10-15 genes, get genetic evidence via OpenTargets_get_evidence_by_datasource 3. Search GWAS Catalog for associations 4. Search OpenTargets GWAS studies 5. Search ClinVar for rare variants 6. For top GWAS genes, check GWAS_search_associations_by_gene

Gene Tracking

Maintain a dictionary of genes found in genomics layer:

genomics_genes = {
    'PSEN1': {'score': 0.87, 'evidence': 'genetic', 'ensembl_id': 'ENSG00000080815', 'layer': 'genomics'},
    'APP': {'score': 0.82, 'evidence': 'genetic', 'ensembl_id': 'ENSG00000142192', 'layer': 'genomics'},
}

---

Phase 2: Transcriptomics Layer Tools

ExpressionAtlas_search_differential:

Input: optional gene (string), condition (string), species (string, default 'homo sapiens')

ExpressionAtlas_search_experiments:

Input: optional gene (string), condition (string), species (string)

expression_atlas_disease_target_score:

Input: efoId (string), pageSize (int, required)

europepmc_disease_target_score:

Input: efoId (string), pageSize (int, required)

HPA_get_rna_expression_by_source (Human Protein Atlas):

Input: gene_name (string), source_type (string: 'tissue', 'blood', 'brain'), source_name (string)
Output: {status, data: {gene_name, source_type, source_name, expression_value, expression_level, expression_unit}}
NOTE: ALL 3 params required. source_type options: 'tissue', 'blood', 'brain', 'cell_line', 'single_cell'

HPA_get_rna_expression_in_specific_tissues:

Input: gene_name (string), tissues (array of strings)

HPA_get_cancer_prognostics_by_gene:

Input: gene_name (string) - for cancer context

HPA_get_subcellular_location:

Input: gene_name (string)

HPA_search_genes_by_query:

Input: query (string)

Phase 2 Workflow

1. Search Expression Atlas for differential expression studies 2. Get expression-based disease scores 3. Get literature-based disease scores (EuropePMC) 4. For top 10-15 genes from genomics layer, check tissue expression via HPA 5. Check disease-relevant tissue expression patterns 6. For cancer: check prognostic biomarkers

---

Phase 3: Proteomics & Interaction Layer Tools

STRING_get_interaction_partners (primary PPI):

Input: protein_ids (array of strings), species (int, default 9606), confidence_score (float, default 0.4), limit (int, default 20)
Output: {status: 'success', data: [{stringId_A, stringId_B, preferredName_A, preferredName_B, ncbiTaxonId, score, nscore, fscore, pscore, ascore, escore, dscore, tscore}]}
NOTE: protein_ids is an array, NOT string. Gene symbols like ['APOE'] work

STRING_get_network:

Input: protein_ids (array), species (int), confidence_score (float)

STRING_functional_enrichment:

Input: protein_ids (array), species (int)

STRING_ppi_enrichment:

Input: protein_ids (array), species (int)

intact_get_interactions:

Input: identifier (string - UniProt ID or gene name)

intact_search_interactions:

Input: query (string), first (int, default 0), max (int, default 25)

HPA_get_protein_interactions_by_gene:

Input: gene_name (string)
Output: {gene, interactions, interactor_count, interactors: [...]}

humanbase_ppi_analysis:

Input: gene_list (array), tissue (string), max_node (int), interaction (string), string_mode (bool)
NOTE: ALL params required. interaction options: 'coexpression', 'interaction', 'coexpression_and_interaction'

Phase 3 Workflow

1. Take top 15-20 genes from genomics + transcriptomics layers 2. Query STRING for interaction partners of each gene 3. Build composite PPI network using STRING_get_network 4. Test PPI enrichment (are genes more connected than random?) 5. Get functional enrichment from STRING 6. For disease-relevant tissue, get tissue-specific network (HumanBase) 7. Identify hub genes (highest degree centrality) 8. Check IntAct for experimentally validated interactions

Hub Gene Analysis

Degree: Number of interaction partners
Betweenness: Number of shortest paths through node
Hub score: Genes with degree > mean + 1 SD are hubs

---

Phase 4: Pathway & Network Layer Tools

enrichr_gene_enrichment_analysis (primary enrichment):

Input: gene_list (array, min 2), libs (array of library names)
Output: {status: 'success', data: '{...JSON string...}'}
Key libraries: ['KEGG_2021_Human'], ['Reactome_2022'], ['WikiPathway_2023_Human'], ['GO_Biological_Process_2023'], ['GO_Molecular_Function_2023'], ['GO_Cellular_Component_2023']
NOTE: data field is a JSON string, needs parsing. libs is REQUIRED as array

ReactomeAnalysis_pathway_enrichment:

Input: identifiers (string - space-separated gene list), optional page_size (int, default 20), include_disease (bool), projection (bool)
Output: {data: {token, analysis_type, pathways_found, pathways: [{pathway_id, name, species, is_disease, is_lowest_level, entities_found, entities_total, entities_ratio, p_value, fdr, reactions_found, reactions_total}]}}

Reactome_map_uniprot_to_pathways:

Input: id (string - UniProt accession)

Reactome_get_pathway / Reactome_get_pathway_reactions:

Input: stId (string, e.g., 'R-HSA-73817')

kegg_search_pathway:

Input: keyword (string)

kegg_get_pathway_info:

Input: pathway_id (string, e.g., 'hsa04930')

WikiPathways_search:

Input: query (string), optional organism (string, e.g., 'Homo sapiens')

Phase 4 Workflow

1. Collect all genes from genomics + transcriptomics layers (top 20-30) 2. Run Enrichr enrichment for KEGG, Reactome, WikiPathways 3. Run ReactomeAnalysis for detailed Reactome enrichment with p-values 4. Search KEGG for disease-specific pathways 5. Search WikiPathways for disease pathways 6. For top Reactome pathways, get detailed reactions 7. Identify cross-pathway connections (genes in multiple pathways)

---

Phase 5: Gene Ontology & Functional Annotation Tools

enrichr_gene_enrichment_analysis (GO enrichment):

Use with libs=['GO_Biological_Process_2023'] for BP
Use with libs=['GO_Molecular_Function_2023'] for MF
Use with libs=['GO_Cellular_Component_2023'] for CC

GO_get_annotations_for_gene:

Input: gene_id (string - gene symbol or UniProt ID)

GO_search_terms:

Input: query (string)

QuickGO_annotations_by_gene:

Input: gene_product_id (string, e.g., 'UniProtKB:P02649'), optional aspect ('biological_process', 'molecular_function', 'cellular_component'), taxon_id (int: 9606), limit (int: 25)

OpenTargets_get_target_gene_ontology_by_ensemblID:

Input: ensemblId (string)

Phase 5 Workflow

1. Run Enrichr GO enrichment for all 3 aspects using combined gene list 2. For top 5 genes, get detailed GO annotations from QuickGO 3. For top genes, get OpenTargets GO terms 4. Summarize key biological processes, molecular functions, cellular components

---

Phase 6: Therapeutic Landscape Tools

OpenTargets_get_associated_drugs_by_disease_efoId (primary):

Input: efoId (string), size (int, REQUIRED - use 100)
Output: {data: {disease: {knownDrugs: {count, rows: [{drug: {id, name, tradeNames, maximumClinicalTrialPhase, isApproved, hasBeenWithdrawn}, phase, mechanismOfAction, target: {id, approvedSymbol}, disease: {id, name}, urls: [{url, name}]}]}}}}

OpenTargets_get_target_tractability_by_ensemblID:

Input: ensemblId (string)

OpenTargets_get_associated_drugs_by_target_ensemblID:

Input: ensemblId (string), size (int, REQUIRED)

search_clinical_trials:

Input: query_term (string, REQUIRED), optional condition (string), intervention (string), pageSize (int, default 10)
NOTE: query_term is REQUIRED even if condition is provided

OpenTargets_get_drug_mechanisms_of_action_by_chemblId:

Input: chemblId (string)

Phase 6 Workflow

1. Get all drugs for disease from OpenTargets 2. For top disease-associated genes, check tractability 3. For top genes with no approved drugs, identify repurposing candidates 4. Search clinical trials for disease 5. For top approved drugs, get mechanism of action

Drug Tracking

drug_targets = {
    'PSEN1': {'drugs': ['Semagacestat'], 'tractability': 'small_molecule', 'clinical_phase': 3},
    'ACHE': {'drugs': ['Donepezil', 'Galantamine'], 'tractability': 'small_molecule', 'clinical_phase': 4},
}

---

Tool Parameter Quick Reference

Tool	Key Parameters	Notes
`OpenTargets_get_disease_id_description_by_name`	`diseaseName`	Primary disambiguation
`OSL_get_efo_id_by_disease_name`	`disease`	Secondary disambiguation
`OpenTargets_get_associated_targets_by_disease_efoId`	`efoId`	Returns top 25 genes
`OpenTargets_get_evidence_by_datasource`	`efoId`, `ensemblId`, `datasourceIds[]`, `size`	Per-gene evidence
`OpenTargets_search_gwas_studies_by_disease`	`diseaseIds[]`, `size`	GWAS studies
`gwas_search_associations`	`disease_trait`, `size`	GWAS Catalog
`ClinVar_search_variants`	`condition` or `gene`, `max_results`	Rare variants
`ExpressionAtlas_search_differential`	`condition`, `species`	DEGs
`expression_atlas_disease_target_score`	`efoId`, `pageSize` (REQUIRED)	Expression scores
`europepmc_disease_target_score`	`efoId`, `pageSize` (REQUIRED)	Literature scores
`HPA_get_rna_expression_by_source`	`gene_name`, `source_type`, `source_name` (ALL REQUIRED)	Tissue expression
`STRING_get_interaction_partners`	`protein_ids[]`, `species` (9606), `limit`	PPI partners
`STRING_get_network`	`protein_ids[]`, `species`	PPI network
`STRING_functional_enrichment`	`protein_ids[]`, `species`	Functional enrichment
`STRING_ppi_enrichment`	`protein_ids[]`, `species`	Network significance
`intact_search_interactions`	`query`, `max`	Experimental PPIs
`humanbase_ppi_analysis`	`gene_list[]`, `tissue`, `max_node`, `interaction`, `string_mode` (ALL REQ)	Tissue PPI
`enrichr_gene_enrichment_analysis`	`gene_list[]`, `libs[]` (BOTH REQUIRED)	Pathway/GO enrichment
`ReactomeAnalysis_pathway_enrichment`	`identifiers` (space-sep string)	Reactome enrichment
`Reactome_map_uniprot_to_pathways`	`id` (UniProt accession)	Protein-pathway mapping
`kegg_search_pathway`	`keyword`	KEGG pathway search
`WikiPathways_search`	`query`, `organism`	WikiPathways search
`GO_get_annotations_for_gene`	`gene_id`	GO annotations
`QuickGO_annotations_by_gene`	`gene_product_id` (e.g., 'UniProtKB:P02649')	Detailed GO
`OpenTargets_get_associated_drugs_by_disease_efoId`	`efoId`, `size` (REQUIRED)	Disease drugs
`OpenTargets_get_target_tractability_by_ensemblID`	`ensemblId`	Druggability
`search_clinical_trials`	`query_term` (REQUIRED), `condition`, `pageSize`	Clinical trials
`PubMed_search_articles`	`query`, `limit`	Literature
`ensembl_lookup_gene`	`gene_id`, `species` ('homo_sapiens' REQUIRED)	Gene lookup
`MyGene_query_genes`	`query`, `species`, `fields`, `size`	Gene info
`OpenTargets_get_similar_entities_by_disease_efoId`	`efoId`, `threshold`, `size` (ALL REQUIRED)	Similar diseases

Use Patterns, Edge Cases & Fallback Strategies

---

Common Use Patterns

1. Comprehensive Disease Profiling

User: "Characterize Alzheimer's disease across omics layers"
-> Run all 8 phases
-> Produce full multi-omics report

2. Therapeutic Target Discovery

User: "What are druggable targets for rheumatoid arthritis?"
-> Emphasize Phase 1 (genomics), Phase 6 (therapeutics), Phase 7 (integration)
-> Focus on tractability and clinical precedent

3. Biomarker Identification

User: "Find diagnostic biomarkers for pancreatic cancer"
-> Emphasize Phase 2 (transcriptomics), Phase 3 (proteomics), Phase 7 (biomarkers)
-> Focus on tissue-specific expression and diagnostic potential

4. Mechanism Elucidation

User: "What pathways are dysregulated in Crohn's disease?"
-> Emphasize Phase 4 (pathways), Phase 5 (GO), Phase 7 (mechanistic hypotheses)
-> Focus on pathway enrichment and cross-pathway connections

5. Drug Repurposing

User: "What existing drugs could be repurposed for ALS?"
-> Emphasize Phase 1 (genetics), Phase 6 (therapeutic landscape), Phase 7 (repurposing)
-> Focus on drugs targeting disease-associated genes

6. Systems Biology

User: "What are the hub genes and key pathways in type 2 diabetes?"
-> Emphasize Phase 3 (PPI network), Phase 4 (pathways), Phase 7 (network analysis)
-> Focus on hub genes and network modules

---

Edge Case Handling

Rare Diseases (limited data)

Genomics layer may dominate (single gene)
Limited GWAS data (monogenic)
Focus on ClinVar variants, pathway consequences
Confidence score will be lower (less cross-layer data)

Common Diseases (overwhelming data)

Thousands of GWAS associations
Prioritize by effect size and significance
Focus on top 20-30 genes for downstream analysis
Use strict significance thresholds (p < 5e-8)

Cancer

Include somatic mutations (if CIViC/cBioPortal available)
Check cancer prognostics via HPA
Include tumor-specific expression patterns
Clinical trial landscape may be extensive

Monogenic Diseases

Single gene dominates
ClinVar/OMIM evidence is primary
Pathway analysis reveals downstream effects
Therapeutic landscape may be limited (gene therapy, enzyme replacement)

Polygenic Diseases

Many weak genetic signals
GWAS provides the gene list
Pathway enrichment reveals convergent biology
Network analysis identifies hub genes

Tissue Ambiguity

Diseases affecting multiple tissues
Query HPA for all relevant tissues
Compare tissue-specific expression patterns
Use tissue context from disease ontology

---

Fallback Strategies

If disease name not found

1. Try synonyms 2. Try broader disease category 3. Try OMIM/UMLS ID mapping 4. Report disambiguation failure and ask user

If no GWAS data

1. Check ClinVar for rare variants 2. Use OpenTargets genetic evidence 3. Note in report as "Limited genetic data" 4. Adjust confidence score accordingly

If no expression data

1. Try different disease name/synonym 2. Check HPA for individual gene expression 3. Use OpenTargets expression evidence 4. Note as "Limited transcriptomics data"

If no pathway enrichment

1. Reduce gene list stringency 2. Try different pathway databases 3. Map individual genes to pathways via Reactome 4. Note as "No significant pathway enrichment"

If no drugs found

1. Check if disease is rare/orphan 2. Look for drugs targeting individual genes 3. Check clinical trials for investigational therapies 4. Note as "No approved drugs - novel therapeutic opportunity"

Related skills

Microsoft FoundryDeploy, evaluate, and continuously improve Microsoft Foundry agents from a single agent interface.490k1.3k

Ai Research ReproductionOrchestrate trustworthy, auditable reproduction of deep learning repositories directly from their READMEs.176k511

Run TrainSafely execute selected deep learning training commands with standardized evidence capture.176k511

Explore RunSafely run isolated exploratory experiments with clear recording and conservative selection before committing changes.176k511

Paper Context ResolverFetch precise reproduction-critical details like dataset splits, preprocessing steps, or evaluation protocols from the original academic paper when the repo README leav141k511

Repo Intake And PlanScan unfamiliar AI research repositories and receive a minimal, trustworthy reproduction target before investing significant time.140k511

How it compares

Choose tooluniverse-multiomic-disease-characterization for full cross-omics disease reports; use tooluniverse-disease-research for lighter disease overviews or GWAS-specific skills when genetics alone is sufficient.

FAQ

How many phases does tooluniverse-multiomic-disease-characterization run?

tooluniverse-multiomic-disease-characterization executes nine phases from disease disambiguation through genomics, transcriptomics, proteomics/PPI, pathways, GO annotation, therapeutics, cross-layer integration, and report finalization with a 0–100 confidence score.

What confidence scoring does tooluniverse-multiomic-disease-characterization use?

tooluniverse-multiomic-disease-characterization grades evidence T1 (human/clinical) through T4 (computational) and calculates a Multi-Omics Confidence Score from 0–100 using the rubric in integration-scoring.md across integrated molecular layers.

Which expression data does tooluniverse-multiomic-disease-characterization query?

tooluniverse-multiomic-disease-characterization uses GTEx_get_expression_summary for baseline expression across 54 tissues plus Expression Atlas and Human Protein Atlas tools to check disease-relevant tissue expression for top genetic candidates.

Data Science & MLanalyticspipelinesetl