Tooluniverse Target Research

Comprehensive Target Intelligence Gatherer

Gather complete target intelligence by exploring 9 parallel research paths. Supports targets identified by gene symbol, UniProt accession, Ensembl ID, or gene name.

KEY PRINCIPLES: 1. Report-first approach - Create report file FIRST, then populate progressively 2. Tool parameter verification - Verify params via get_tool_info before calling unfamiliar tools 3. Evidence grading - Grade all claims by evidence strength (T1-T4) 4. Citation requirements - Every fact must have inline source attribution 5. Mandatory completeness - All sections must exist with data minimums or explicit "No data" notes 6. Disambiguation first - Resolve all identifiers before research 7. Negative results documented - "No drugs found" is data; empty sections are failures 8. Collision-aware literature search - Detect and filter naming collisions 9. English-first queries - Always use English terms in tool calls, even if the user writes in another language. Translate gene names, disease names, and search terms to English. Only try original-language terms as a fallback if English returns no results. Respond in the user's language

---

LOOK UP, DON'T GUESS

When asked about a specific protein or gene target, look it up in UniProt/Ensembl/OpenTargets BEFORE reasoning about it. Verify the gene name, function, and disease associations from databases. When you're not sure about a fact, your first instinct should be to SEARCH for it using tools, not to reason harder from memory.

---

When to Use This Skill

Apply when users:

• Ask about a drug target, protein, or gene
• Need target validation or assessment
• Request druggability analysis
• Want comprehensive target profiling
• Ask "what do we know about [target]?"
• Need target-disease associations
• Request safety profile for a target

When NOT to use: Simple protein lookup, drug-only queries, disease-centric queries, sequence retrieval, structure download — use specialized skills instead.

---

Target Evaluation Reasoning Framework

Evaluating a drug target requires reasoning across four interconnected questions. Answer all four before forming a recommendation.

1. Is there genetic evidence linking this target to the disease? Genetic evidence is the strongest predictor of drug success — targets with human genetic support have approximately twice the clinical success rate as those without (Nelson et al. 2015). Ask: Are there GWAS associations connecting this gene to the disease? Do rare loss-of-function or gain-of-function variants cause or protect against the disease? Does the mouse knockout phenotype match the human disease (from OpenTargets mouse models)? OpenTargets assigns genetic evidence scores; a score > 0.7 indicates strong support. ClinVar rare variant evidence and DisGeNET curated gene-disease association scores add complementary layers. A target with no genetic link to the disease of interest carries a fundamental validation risk that cannot be resolved by downstream data.

2. Is the target druggable? Druggability has two components: structural accessibility and prior chemical matter. Structural accessibility means the target has a binding pocket where a small molecule or biologic can engage — surface-exposed receptors, enzymes with well-defined active sites, and protein-protein interaction interfaces with hot spots are tractable. Intrinsically disordered proteins and transcription factors with flat, featureless binding surfaces are typically harder. Pharos TDL classification provides a tiered assessment: Tclin (approved drug), Tchem (known active compounds), Tbio (biological function known but no drugs), Tdark (poorly characterized). If ChEMBL or BindingDB have compounds with IC50 < 1μM, the target is chemically tractable. Chemical probes (from OpenTargets chemical probes endpoint) confirm a target can be modulated, which is distinct from drug-like compounds. For GPCRs, check GPCRdb for curated agonists and antagonists.

3. Is the target safe to modulate? Safety concerns arise from two sources. First, on-target effects: if the target is essential in normal tissues (mouse KO is lethal, or gnomAD pLI is high / LOEUF is low), full inhibition will produce toxicity — the question becomes whether a partial agonist or tissue-targeted delivery can provide a therapeutic window. Second, off-target effects: does the gene have family members that could be inadvertently hit? The OpenTargets safety profile aggregates known toxicity annotations, and DepMap essentiality scores tell you which cancer cell lines require this gene for survival (useful but not directly translatable to normal tissues). Expression specificity matters: a target expressed only in the disease-relevant tissue is far safer than one expressed ubiquitously in critical organs (heart, kidney, brain).

4. What is the competitive landscape? A target with approved drugs may already be validated but competitive; a target with clinical-stage programs from competitors establishes feasibility while creating IP barriers. An entirely novel target with no drug history requires more extensive internal validation. Assess: number of ChEMBL bioactivity records (chemical matter depth), approved drugs from OpenTargets drug associations, and literature activity trends (recent paper count and key research groups). A dark target (Tdark) with strong genetic evidence but no chemical matter is a high-risk, high-reward opportunity.

Synthesizing the four dimensions: The ideal target has strong genetic evidence (GWAS + rare variant), a tractable binding site (Tclin or Tchem), acceptable safety profile (tissue-specific expression, non-lethal KO), and manageable competition. Gaps in any dimension represent validation tasks, not disqualifiers — but they must be acknowledged. A target with perfect druggability but no genetic link to disease is a tractability exercise, not a validated therapeutic hypothesis.

---

Phase 0: Tool Parameter Verification (CRITICAL)

BEFORE calling ANY tool for the first time, verify its parameters:

tool_info = tu.tools.get_tool_info(tool_name="Reactome_map_uniprot_to_pathways")
# Reveals: takes `id` not `uniprot_id`

Known parameter corrections:

• Reactome_map_uniprot_to_pathways: param is id (not uniprot_id)
• ensembl_get_xrefs: param is id (not gene_id)
• GTEx_get_median_gene_expression: requires gencode_id + operation="median"; try versioned Ensembl ID if empty
• OpenTargets_*: param is ensemblId (camelCase, not ensemblID)
• STRING_get_protein_interactions: takes protein_ids (list) + species
• intact_get_interactions: takes identifier (UniProt accession, not gene symbol)

---

Critical Workflow Requirements

Report-First (MANDATORY): Create [TARGET]_target_report.md with all section headers and [Researching...] placeholders before starting research. Update progressively. Do not show raw tool outputs to the user.

Evidence Grading (MANDATORY): Grade every claim T1-T4. T1 = clinical/genetic data; T2 = curated databases or multiple studies; T3 = computational or single study; T4 = annotation or catalog entry.

---

Core Strategy: 9 Research Paths

Target Query (e.g., "EGFR" or "P00533")
|
+- IDENTIFIER RESOLUTION (always first)
|   +- Check if GPCR -> GPCRdb_get_protein
|
+- PATH 0: Open Targets Foundation (ALWAYS FIRST - fills gaps in all other paths)
|
+- PATH 1: Core Identity (names, IDs, sequence, organism)
|   +- InterProScan_scan_sequence for novel domain prediction
+- PATH 2: Structure & Domains (3D structure, domains, binding sites)
|   +- If GPCR: GPCRdb_get_structures (active/inactive states)
+- PATH 3: Function & Pathways (GO terms, pathways, biological role)
+- PATH 4: Protein Interactions (PPI network, complexes)
+- PATH 5: Expression Profile (tissue expression, single-cell)
+- PATH 6: Variants & Disease (mutations, clinical significance)
|   +- DisGeNET_search_gene for curated gene-disease associations
+- PATH 7: Drug Interactions (known drugs, druggability, safety)
|   +- Pharos_get_target for TDL classification (Tclin/Tchem/Tbio/Tdark)
|   +- BindingDB_get_ligands_by_uniprot for known ligands
|   +- PubChem_search_assays_by_target_gene for HTS data
|   +- If GPCR: GPCRdb_get_ligands (curated agonists/antagonists)
|   +- DepMap_get_gene_dependencies for target essentiality
+- PATH 8: Literature & Research (publications, trends)

For detailed code implementations of each path, see IMPLEMENTATION.md.

---

Identifier Resolution (Phase 1)

Resolve ALL identifiers before any research path. Required IDs:

• UniProt accession (for protein data, structure, interactions)
• Ensembl gene ID + versioned ID (for Open Targets, GTEx)
• Gene symbol (for DGIdb, gnomAD, literature)
• Entrez gene ID (for KEGG, MyGene)
• ChEMBL target ID (for bioactivity)
• Synonyms/full name (for collision-aware literature search)

After resolution, check if target is a GPCR via GPCRdb_get_protein. See IMPLEMENTATION.md for resolution and GPCR detection code.

---

PATH 0: Open Targets Foundation (ALWAYS FIRST)

Run OpenTargets endpoints first to populate baseline data before specialized queries:

• OpenTargets_get_diseases_phenotypes_by_target_ensembl → disease associations (Section 8)
• OpenTargets_get_target_tractability_by_ensemblID → druggability assessment (Section 9)
• OpenTargets_get_target_safety_profile_by_ensemblID → safety liabilities (Section 10)
• OpenTargets_get_target_interactions_by_ensemblID → PPI network (Section 6)
• OpenTargets_get_target_gene_ontology_by_ensemblID → GO annotations (Section 5)
• OpenTargets_get_publications_by_target_ensemblID → literature (Section 11)
• OpenTargets_get_biological_mouse_models_by_ensemblID → mouse KO phenotypes (Sections 8/10)
• OpenTargets_get_chemical_probes_by_target_ensemblID → chemical probes (Section 9)
• OpenTargets_get_associated_drugs_by_target_ensemblID → known drugs (Section 9)

---

PATH 1: Core Identity

Tools: UniProt_get_entry_by_accession, UniProt_get_function_by_accession, UniProt_get_recommended_name_by_accession, UniProt_get_alternative_names_by_accession, UniProt_get_subcellular_location_by_accession, MyGene_get_gene_annotation

Populates: Sections 2 (Identifiers), 3 (Basic Information)

---

PATH 2: Structure & Domains

Use 3-step structure search chain (do NOT rely solely on PDB text search): 1. UniProt PDB cross-references (most reliable) 2. Sequence-based PDB search (catches missing annotations) 3. Domain-based search (for multi-domain proteins) 4. AlphaFold (always check)

Tools: UniProt_get_entry_by_accession (PDB xrefs), RCSBData_get_entry, PDB_search_similar_structures, alphafold_get_prediction, InterPro_get_protein_domains, UniProt_get_ptm_processing_by_accession

GPCR targets: Also query GPCRdb_get_structures for active/inactive state data.

Populates: Section 4 (Structural Biology)

---

PATH 3: Function & Pathways

Tools: GO_get_annotations_for_gene, Reactome_map_uniprot_to_pathways, kegg_get_gene_info, WikiPathways_search, enrichr_gene_enrichment_analysis

Populates: Section 5 (Function & Pathways)

---

PATH 4: Protein Interactions

Tools: STRING_get_protein_interactions, intact_get_interactions, intact_get_complex_details, BioGRID_get_interactions, HPA_get_protein_interactions_by_gene

Minimum: 20 interactors OR documented explanation.

Populates: Section 6 (Protein-Protein Interactions)

---

PATH 5: Expression Profile

GTEx with versioned ID fallback + HPA as backup.

Tools: GTEx_get_median_gene_expression, HPA_get_rna_expression_by_source, HPA_get_comprehensive_gene_details_by_ensembl_id, HPA_get_subcellular_location, HPA_get_cancer_prognostics_by_gene, HPA_get_comparative_expression_by_gene_and_cellline, CELLxGENE_get_expression_data

Reasoning: Expression specificity directly informs safety. Note whether expression is enriched in the disease-relevant tissue vs. critical organs. Ubiquitous essential expression narrows the therapeutic window.

Populates: Section 7 (Expression Profile)

---

PATH 6: Variants & Disease

Separate SNVs from CNVs in ClinVar results. Integrate DisGeNET for curated gene-disease association scores.

Tools: gnomad_get_gene_constraints, ClinVar_search_variants, OpenTargets_get_diseases_phenotypes_by_target_ensembl, DisGeNET_search_gene, civic_get_variants_by_gene, cBioPortal_get_mutations

Required constraint scores: pLI (probability of loss-of-function intolerance), LOEUF (loss-of-function observed/expected upper bound), missense Z-score, pRec (recessive probability). High pLI (> 0.9) or low LOEUF (< 0.35) indicates the gene is intolerant to loss-of-function — a major safety flag for inhibitory therapeutic strategies.

Populates: Section 8 (Genetic Variation & Disease)

---

PATH 7: Druggability & Target Validation

Tools: OpenTargets_get_target_tractability_by_ensemblID, DGIdb_get_gene_druggability, DGIdb_get_drug_gene_interactions, ChEMBL_search_targets, ChEMBL_get_target_activities, Pharos_get_target, BindingDB_get_ligands_by_uniprot, PubChem_search_assays_by_target_gene, DepMap_get_gene_dependencies, OpenTargets_get_target_safety_profile_by_ensemblID, OpenTargets_get_biological_mouse_models_by_ensemblID

GPCR targets: Also query GPCRdb_get_ligands.

Reasoning: Pharos TDL tells you where the target sits in the knowledge landscape. BindingDB Ki/IC50/Kd values tell you whether the target has been demonstrated tractable experimentally. DepMap essentiality tells you whether cancer cells require this gene (proxy for toxicity risk, not a definitive answer).

Populates: Sections 9 (Druggability), 10 (Safety), 12 (Competitive Landscape)

---

PATH 8: Literature & Research (Collision-Aware)

1. Detect collisions - Check if gene symbol has non-biological meanings 2. Build seed queries - Symbol in title with bio context, full name, UniProt accession 3. Apply collision filter - Add NOT terms for off-topic meanings 4. Expand via citations - For sparse targets (<30 papers), use citation network 5. Classify by evidence tier - T1-T4 based on title/abstract keywords

Tools: PubMed_search_articles, PubMed_get_related, EuropePMC_search_articles, EuropePMC_get_citations, PubTator3_LiteratureSearch, OpenTargets_get_publications_by_target_ensemblID

Populates: Section 11 (Literature & Research Landscape)

---

Retry Logic & Fallback Chains

• ChEMBL_get_target_activities fails → GtoPdb_search_ligands → OpenTargets drugs
• intact_get_interactions fails → STRING_get_protein_interactions → OpenTargets interactions
• GO_get_annotations_for_gene fails → OpenTargets GO → MyGene GO
• GTEx_get_median_gene_expression fails → HPA_get_rna_expression_by_source → document as unavailable
• gnomad_get_gene_constraints fails → OpenTargets constraint endpoint
• DGIdb_get_drug_gene_interactions fails → OpenTargets drugs → GtoPdb_search_ligands

NEVER silently skip failed tools. Always document failures and fallbacks in the report.

---

Completeness Audit (REQUIRED before finalizing)

Before finalizing any report:

• Data minimums met for PPIs, expression, diseases, constraints, druggability
• Negative results documented explicitly
• T1-T4 grades in Executive Summary, Disease Associations, Key Papers, Recommendations
• Every data point has source attribution

---

Report Template

Create [TARGET]_target_report.md with all 15 sections initialized. See REPORT_FORMAT.md for the full template.

## 1. Executive Summary          ## 9. Druggability & Pharmacology
## 2. Target Identifiers         ## 10. Safety Profile
## 3. Basic Information          ## 11. Literature & Research
## 4. Structural Biology         ## 12. Competitive Landscape
## 5. Function & Pathways        ## 13. Summary & Recommendations
## 6. Protein-Protein Interactions ## 14. Data Sources & Methodology
## 7. Expression Profile         ## 15. Data Gaps & Limitations
## 8. Genetic Variation & Disease

---

Synthesis: Target Assessment Framework

After completing all 9 PATHs, synthesize findings into a GO/NO-GO recommendation in the Executive Summary. Score each dimension:

• Genetic evidence: Strong (GWAS + rare variant + functional) / Moderate (GWAS or rare variant only) / Weak (expression change only) / None
• Disease association: Based on OpenTargets score (> 0.7 strong, 0.3-0.7 moderate, < 0.3 weak)
• Druggability: Approved drug exists / Tractable (known binding site, chemical probes) / Predicted tractable (structural pocket) / Undruggable
• Safety: Non-essential gene (viable KO, low pLI) / Essential with phenotype / Lethal KO or high pLI / Known toxicity target
• Selectivity: Disease-specific or enriched expression / Ubiquitous / Expressed in critical organs
• Structural data: High-res crystal with ligand / AlphaFold confident (pLDDT > 80) / Homology model / No structural info

Total score guides recommendation: strong target (all dimensions favorable), promising with defined validation tasks (2-3 gaps), speculative (multiple critical gaps), or deprioritize (no genetic link and poor druggability).

---

Reference Files

Evidence Grading & Completeness Audit

Evidence grading system and completeness requirements for target intelligence reports.

Evidence Tiers

Required Evidence Grading Locations

Evidence grades MUST appear in: 1. Executive Summary - Key disease claims graded 2. Section 8.2 Disease Associations - Every disease link graded with source type 3. Section 11 Literature - Key papers table with evidence tier 4. Section 13 Recommendations - Scorecard items reference evidence quality

Per-Section Evidence Summary Format

---
**Evidence Quality for this Section**: Strong
- Mechanistic (T1): 12 papers
- Functional (T2): 8 papers
- Association (T3): 15 papers
- Mention (T4): 23 papers
**Data Gaps**: No CRISPR data; mouse KO phenotypes limited
---

Citation Format

Every piece of information MUST include its source:

EGFR mutations cause lung adenocarcinoma [three stars: PMID:15118125, activating mutations
in patients]. *Source: ClinVar, CIViC*

ClinVar SNV vs CNV Separation

Always separate single nucleotide variants from copy number variants:

### 8.3 Clinical Variants (ClinVar)

#### Single Nucleotide Variants (SNVs)
| Variant | Clinical Significance | Condition | Review Status | PMID |
|---------|----------------------|-----------|---------------|------|
| p.L858R | Pathogenic | Lung cancer | 4 stars | 15118125 |

**Total Pathogenic SNVs**: 47

#### Copy Number Variants (CNVs) - Reported Separately
| Type | Region | Clinical Significance | Frequency |
|------|--------|----------------------|-----------|
| Amplification | 7p11.2 | Pathogenic | Common in cancer |

*Note: CNV data separated as it represents different mutation mechanism*

DisGeNET Evidence Tier Assignment

• DisGeNET Score >= 0.7 -> Consider T2 evidence (multiple validated sources)
• DisGeNET Score 0.4-0.7 -> Consider T3 evidence
• DisGeNET Score < 0.4 -> T4 evidence only

---

Minimum Data Requirements (Enforced)

Post-Run Completeness Audit

Before finalizing the report, run this checklist:

Data Minimums Check

• [ ] PPIs: >= 20 interactors OR explanation why fewer
• [ ] Expression: Top 10 tissues with values OR explicit "unavailable"
• [ ] Diseases: Top 10 associations with scores OR "no associations"
• [ ] Constraints: All 4 scores (pLI, LOEUF, missense Z, pRec) OR "unavailable"
• [ ] Druggability: All modalities assessed; probes + drugs listed OR "none"

Negative Results Documented

• [ ] Empty tool results noted explicitly (not left blank)
• [ ] Failed tools with fallbacks documented
• [ ] "No data" sections have implications noted

Evidence Quality

• [ ] T1-T4 grades in Executive Summary disease claims
• [ ] T1-T4 grades in Disease Associations table
• [ ] Key papers table has evidence tiers
• [ ] Per-section evidence summaries included

Source Attribution

• [ ] Every data point has source tool/database cited
• [ ] Section-end source summaries present

Data Gap Table (Required if minimums not met)

## 15. Data Gaps & Limitations

| Section | Expected Data | Actual | Reason | Alternative Source |
|---------|---------------|--------|--------|-------------------|
| 6. PPIs | >= 20 interactors | 8 | Novel target, limited studies | Literature review needed |
| 7. Expression | GTEx TPM | None | Versioned ID not recognized | See HPA data |
| 9. Probes | Chemical probes | None | No validated probes exist | Consider tool compound dev |

**Recommendations for Data Gaps**:
1. For PPIs: Query BioGRID with broader parameters; check yeast-2-hybrid studies
2. For Expression: Query GEO directly for tissue-specific datasets

Target Intelligence Examples

Detailed examples showing multi-step workflows for comprehensive target analysis.

Example 1: Complete EGFR Target Profile

Query: "Tell me everything about EGFR as a drug target"

Step 1: Resolve Identifiers

from tooluniverse import ToolUniverse

tu = ToolUniverse(use_cache=True)
tu.load_tools()

# Resolve EGFR to all IDs
# Search UniProt for human EGFR
search_result = tu.tools.UniProt_search(
    query='gene:EGFR AND organism_id:9606',
    limit=1
)
uniprot_id = search_result['results'][0]['primaryAccession']  # P00533

# Map to Ensembl
mapping = tu.tools.UniProt_id_mapping(
    ids=['P00533'],
    from_db='UniProtKB_AC-ID',
    to_db='Ensembl'
)
ensembl_id = mapping['results'][0]['to']  # ENSG00000146648

ids = {
    'symbol': 'EGFR',
    'uniprot': 'P00533',
    'ensembl': 'ENSG00000146648'
}

Step 2: Core Identity (PATH 1)

# Get full UniProt entry
entry = tu.tools.UniProt_get_entry_by_accession(accession='P00533')

# Extract key info
identity = {
    'name': entry.get('proteinDescription', {}).get('recommendedName', {}).get('fullName', {}).get('value'),
    'length': entry.get('sequence', {}).get('length'),
    'organism': entry.get('organism', {}).get('scientificName'),
    'function': tu.tools.UniProt_get_function_by_accession(accession='P00533')
}

# Get gene annotation
gene_info = tu.tools.MyGene_get_gene_annotation(
    gene_id='1956',  # Entrez ID for EGFR
    fields='symbol,name,summary,alias,genomic_pos'
)

Output:

Name: Epidermal growth factor receptor
Symbol: EGFR
Length: 1210 amino acids
Organism: Homo sapiens
Function: Receptor tyrosine kinase binding ligands of the EGF family...

Step 3: Structure & Domains (PATH 2)

# Get PDB structures from UniProt entry
pdb_refs = [xref for xref in entry.get('uniProtKBCrossReferences', []) 
            if xref.get('database') == 'PDB']

# Get details for best structure
best_pdb = '1M17'  # Example: EGFR kinase domain
pdb_info = tu.tools.get_protein_metadata_by_pdb_id(pdb_id='1M17')

# Get AlphaFold prediction
alphafold = tu.tools.alphafold_get_prediction(qualifier='P00533')

# Get domain architecture
domains = tu.tools.InterPro_get_protein_domains(protein_id='P00533')

# Get PTMs and active sites
ptms = tu.tools.UniProt_get_ptm_processing_by_accession(accession='P00533')

Output:

PDB Structures: 150+ entries
Best Resolution: 1.8Å (1M17)
AlphaFold: Available, high confidence
Key Domains:
  - Receptor L domain (EGF binding)
  - Furin-like domain
  - Growth factor receptor domain
  - Protein kinase domain
PTMs: Multiple phosphorylation sites (Y992, Y1068, Y1086...)

Step 4: Function & Pathways (PATH 3)

# GO annotations
go_terms = tu.tools.GO_get_annotations_for_gene(gene_id='UniProtKB:P00533')

# Reactome pathways
reactome_pathways = tu.tools.Reactome_map_uniprot_to_pathways(id='P00533')

# KEGG pathways
kegg_info = tu.tools.kegg_get_gene_info(gene_id='hsa:1956')

# Open Targets GO
ot_go = tu.tools.OpenTargets_get_target_gene_ontology_by_ensemblID(
    ensemblID='ENSG00000146648'
)

Output:

GO Biological Process:
  - Signal transduction (GO:0007165)
  - Cell proliferation (GO:0008283)
  - MAPK cascade (GO:0000165)
  
GO Molecular Function:
  - Protein tyrosine kinase activity (GO:0004713)
  - ATP binding (GO:0005524)
  - Receptor binding (GO:0005102)

Key Pathways:
  - EGFR signaling pathway (Reactome)
  - PI3K-Akt signaling (KEGG hsa04151)
  - MAPK signaling (KEGG hsa04010)
  - ErbB signaling pathway (KEGG hsa04012)

Step 5: Protein Interactions (PATH 4)

# STRING interactions
string_ppi = tu.tools.STRING_get_protein_interactions(
    protein_ids=['EGFR'],
    species=9606,
    confidence_score=0.9,
    limit=50
)

# IntAct experimental interactions
intact_ppi = tu.tools.intact_get_interactions(
    identifier='P00533',
    format='json'
)

# Open Targets interactions
ot_ppi = tu.tools.OpenTargets_get_target_interactions_by_ensemblID(
    ensemblID='ENSG00000146648'
)

Output:

Top STRING Interactors (score > 0.9):
  1. GRB2 (0.999) - Adapter protein
  2. SHC1 (0.999) - Signal transduction
  3. ERBB2 (0.998) - Receptor family
  4. SRC (0.996) - Kinase
  5. STAT3 (0.994) - Transcription factor

IntAct Complexes:
  - EGFR-GRB2-SOS1 complex
  - EGFR-PI3K complex

Step 6: Expression Profile (PATH 5)

# GTEx expression
gtex = tu.tools.GTEx_get_median_gene_expression(
    gencode_id='ENSG00000146648.11'
)

# HPA expression
hpa = tu.tools.HPA_get_comprehensive_gene_details_by_ensembl_id(
    ensembl_id='ENSG00000146648'
)

# Subcellular location
subcell = tu.tools.HPA_get_subcellular_location(
    ensembl_id='ENSG00000146648'
)

# Cancer prognostics
cancer = tu.tools.HPA_get_cancer_prognostics_by_gene(
    gene_symbol='EGFR'
)

Output:

Top Expression Tissues (GTEx TPM):
  1. Skin (150+ TPM)
  2. Esophagus mucosa (120+ TPM)
  3. Kidney cortex (100+ TPM)
  4. Lung (80+ TPM)

Tissue Specificity: Low (broadly expressed)
Subcellular: Plasma membrane, Cytoplasm

Cancer Relevance:
  - Overexpressed in: NSCLC, Glioblastoma, Colorectal
  - Prognostic: Unfavorable in lung cancer

Step 7: Variants & Disease (PATH 6)

# gnomAD constraint scores
constraints = tu.tools.gnomad_get_gene_constraints(gene_symbol='EGFR')

# UniProt disease variants
disease_vars = tu.tools.UniProt_get_disease_variants_by_accession(accession='P00533')

# ClinVar variants
clinvar = tu.tools.ClinVar_search_variants(gene='EGFR', max_results=100)

# Open Targets disease associations
diseases = tu.tools.OpenTargets_get_diseases_phenotypes_by_target_ensembl(
    ensemblId='ENSG00000146648'
)

Output:

Constraint Scores:
  - pLI: 0.99 (highly loss-of-function intolerant)
  - LOEUF: 0.18
  - Missense Z: 3.5

Disease Associations (Open Targets):
  1. Non-small cell lung carcinoma (0.95)
  2. Glioblastoma multiforme (0.89)
  3. Colorectal cancer (0.82)
  4. Pancreatic cancer (0.75)
  
ClinVar Pathogenic Variants: 45
  - L858R (common activating mutation)
  - T790M (resistance mutation)
  - Exon 19 deletions

Step 8: Drug Interactions (PATH 7)

# Open Targets tractability
tractability = tu.tools.OpenTargets_get_target_tractability_by_ensemblID(
    ensemblID='ENSG00000146648'
)

# DGIdb druggability
druggability = tu.tools.DGIdb_get_gene_druggability(genes=['EGFR'])

# Known drugs
drugs = tu.tools.OpenTargets_get_associated_drugs_by_target_ensemblID(
    ensemblID='ENSG00000146648'
)

# ChEMBL bioactivity
# First get ChEMBL target ID
chembl_target = tu.tools.ChEMBL_search_targets(
    pref_name__contains='EGFR',
    organism='Homo sapiens',
    limit=1
)
target_chembl_id = chembl_target['targets'][0]['target_chembl_id']  # CHEMBL203

activities = tu.tools.ChEMBL_get_target_activities(
    target_chembl_id__exact='CHEMBL203',
    limit=100
)

# Safety profile
safety = tu.tools.OpenTargets_get_target_safety_profile_by_ensemblID(
    ensemblID='ENSG00000146648'
)

# Chemical probes
probes = tu.tools.OpenTargets_get_chemical_probes_by_target_ensemblID(
    ensemblID='ENSG00000146648'
)

Output:

Tractability:
  - Small Molecule: HIGH (clinical precedence)
  - Antibody: HIGH (approved antibodies)
  - PROTAC: HIGH (structural data available)
  - Other Modalities: MEDIUM

Approved Drugs:
  1. Erlotinib (Tarceva) - TKI
  2. Gefitinib (Iressa) - TKI
  3. Afatinib (Gilotrif) - TKI
  4. Osimertinib (Tagrisso) - 3rd gen TKI
  5. Cetuximab (Erbitux) - mAb
  6. Panitumumab (Vectibix) - mAb

ChEMBL Activities:
  - 50,000+ bioactivity records
  - Best IC50: 0.5 nM (osimertinib)

Safety Liabilities:
  - Skin toxicity (class effect)
  - Diarrhea (common)
  - Interstitial lung disease (rare)

Chemical Probes: 
  - Gefitinib (SGC probe)

Step 9: Literature (PATH 8)

# PubMed publications
pubmed_total = tu.tools.PubMed_search_articles(
    query='EGFR[Gene Name]',
    limit=0  # Just count
)

pubmed_recent = tu.tools.PubMed_search_articles(
    query='EGFR[Gene Name] AND "2024"[Date - Publication]',
    limit=0
)

pubmed_drug = tu.tools.PubMed_search_articles(
    query='EGFR AND drug AND cancer',
    limit=10
)

# Open Targets publications
ot_pubs = tu.tools.OpenTargets_get_publications_by_target_ensemblID(
    ensemblID='ENSG00000146648',
    size=10
)

Output:

Literature Summary:
  - Total publications: 180,000+
  - Recent (2024): 8,000+
  - Drug-related: 45,000+
  - Trend: Stable (mature, well-studied target)

Recent Focus Areas:
  - Resistance mechanisms
  - Third-generation inhibitors
  - Combination therapies
  - Liquid biopsy for monitoring

Final Synthesized Report

# Target Intelligence Report: EGFR

## Quick Facts
| Property | Value |
|----------|-------|
| Symbol | EGFR |
| UniProt | P00533 |
| Ensembl | ENSG00000146648 |
| Name | Epidermal growth factor receptor |
| Length | 1210 amino acids |
| Organism | Homo sapiens |

## Druggability Assessment
| Modality | Score | Evidence |
|----------|-------|----------|
| Small molecule | ✅ HIGH | 4+ approved TKIs |
| Antibody | ✅ HIGH | 2+ approved mAbs |
| PROTAC | ✅ HIGH | Structural data |

## Summary by Domain

### Identity & Function
- Receptor tyrosine kinase of the ErbB family
- Regulates cell proliferation, survival, differentiation
- Activates RAS-MAPK and PI3K-AKT pathways

### Structure
- 150+ PDB structures available
- Key domains: Kinase (for TKIs), Extracellular (for mAbs)
- AlphaFold model: High confidence

### Expression
- Broadly expressed (skin, lung, kidney highest)
- Overexpressed in multiple cancers
- Subcellular: Plasma membrane

### Variants & Disease
- Highly constrained (pLI=0.99)
- Major cancer associations: NSCLC, glioblastoma, CRC
- Key mutations: L858R, T790M, exon 19 del

### Drugs
- 6+ approved drugs
- 200+ clinical trials
- Well-characterized safety profile (skin toxicity)

### Research
- Mature target (180K+ publications)
- Active research on resistance mechanisms

## Recommendations
1. [HIGH] Excellent target validation - multiple approved therapies
2. [MEDIUM] Consider resistance mutations in drug design
3. [INFO] Extensive structural data for SBDD available

---

Example 2: Novel Target Assessment (KRAS G12C)

Query: "Is KRAS druggable? What's the current state?"

Multi-Step Workflow

from tooluniverse import ToolUniverse
from concurrent.futures import ThreadPoolExecutor

tu = ToolUniverse(use_cache=True)
tu.load_tools()

# Resolve KRAS
ids = {
    'symbol': 'KRAS',
    'uniprot': 'P01116',
    'ensembl': 'ENSG00000133703'
}

# Parallel execution
def assess_druggability():
    results = {}
    
    # 1. Tractability
    results['tractability'] = tu.tools.OpenTargets_get_target_tractability_by_ensemblID(
        ensemblID=ids['ensembl']
    )
    
    # 2. DGIdb assessment
    results['dgidb'] = tu.tools.DGIdb_get_gene_druggability(genes=['KRAS'])
    
    # 3. Known drugs
    results['drugs'] = tu.tools.OpenTargets_get_associated_drugs_by_target_ensemblID(
        ensemblID=ids['ensembl']
    )
    
    # 4. ChEMBL activities
    chembl = tu.tools.ChEMBL_search_targets(
        pref_name__contains='KRAS',
        organism='Homo sapiens',
        limit=1
    )
    if chembl.get('targets'):
        target_id = chembl['targets'][0]['target_chembl_id']
        results['activities'] = tu.tools.ChEMBL_get_target_activities(
            target_chembl_id__exact=target_id,
            limit=50
        )
    
    # 5. Structures
    results['structures'] = tu.tools.alphafold_get_prediction(qualifier='P01116')
    
    # 6. Safety
    results['safety'] = tu.tools.OpenTargets_get_target_safety_profile_by_ensemblID(
        ensemblID=ids['ensembl']
    )
    
    return results

druggability = assess_druggability()

Output:

KRAS Druggability Assessment:

Tractability:
  - Small Molecule: MEDIUM → HIGH (recent breakthroughs!)
  - Previously "undruggable" - now validated
  
Approved Drugs (G12C-specific):
  - Sotorasib (Lumakras) - 2021
  - Adagrasib (Krazati) - 2022

ChEMBL Activities:
  - 5000+ bioactivity records
  - G12C-specific covalent inhibitors

Structure:
  - Multiple G12C-bound structures
  - Switch II pocket (key for covalent inhibitors)

Recent Breakthrough:
  - Covalent inhibitors targeting G12C mutant
  - Switch II pocket discovered as druggable site
  - Active clinical development for other mutations (G12D, G12V)

---

Example 3: Target Comparison

Query: "Compare EGFR vs HER2 as drug targets"

Parallel Analysis

targets = [
    {'symbol': 'EGFR', 'uniprot': 'P00533', 'ensembl': 'ENSG00000146648'},
    {'symbol': 'ERBB2', 'uniprot': 'P04626', 'ensembl': 'ENSG00000141736'}  # HER2
]

def analyze_target(target):
    result = {
        'symbol': target['symbol'],
        'tractability': tu.tools.OpenTargets_get_target_tractability_by_ensemblID(
            ensemblID=target['ensembl']
        ),
        'drugs': tu.tools.OpenTargets_get_associated_drugs_by_target_ensemblID(
            ensemblID=target['ensembl']
        ),
        'diseases': tu.tools.OpenTargets_get_diseases_phenotypes_by_target_ensembl(
            ensemblId=target['ensembl']
        ),
        'safety': tu.tools.OpenTargets_get_target_safety_profile_by_ensemblID(
            ensemblID=target['ensembl']
        ),
        'ppi': tu.tools.STRING_get_protein_interactions(
            protein_ids=[target['symbol']],
            species=9606,
            confidence_score=0.9,
            limit=20
        )
    }
    return result

# Parallel analysis
with ThreadPoolExecutor(max_workers=2) as executor:
    results = list(executor.map(analyze_target, targets))

Comparison Output:

| Property | EGFR | HER2/ERBB2 |
|----------|------|------------|
| **Approved Drugs** | 6+ | 8+ |
| **TKI Tractability** | HIGH | HIGH |
| **mAb Tractability** | HIGH | HIGH |
| **ADC** | N/A | HIGH (T-DM1, T-DXd) |
| **Primary Indication** | NSCLC | Breast cancer |
| **Key Mutations** | L858R, T790M | Amplification |
| **Safety Concerns** | Skin toxicity | Cardiotoxicity |

---

Example 4: Target Validation Pipeline

Query: "Validate CDK4 as a potential drug target for cancer"

Systematic Validation

def validate_target(gene_symbol, disease_area='cancer'):
    """
    Systematic target validation following industry best practices.
    """
    tu = ToolUniverse(use_cache=True)
    tu.load_tools()
    
    validation_results = {}
    
    # 1. GENETIC EVIDENCE
    # Disease associations
    ensembl_id = 'ENSG00000135446'  # CDK4
    
    disease_assoc = tu.tools.OpenTargets_get_diseases_phenotypes_by_target_ensembl(
        ensemblId=ensembl_id
    )
    validation_results['genetic_evidence'] = {
        'disease_associations': disease_assoc,
        'score': 'HIGH' if any(d.get('score', 0) > 0.5 for d in disease_assoc.get('data', [])) else 'LOW'
    }
    
    # Constraint score
    constraint = tu.tools.gnomad_get_gene_constraints(gene_symbol='CDK4')
    validation_results['constraint'] = constraint
    
    # 2. EXPRESSION EVIDENCE
    # Cancer expression
    hpa_cancer = tu.tools.HPA_get_cancer_prognostics_by_gene(gene_symbol='CDK4')
    validation_results['expression'] = hpa_cancer
    
    # 3. FUNCTIONAL EVIDENCE
    # Pathways
    pathways = tu.tools.Reactome_map_uniprot_to_pathways(id='P11802')
    go_terms = tu.tools.GO_get_annotations_for_gene(gene_id='UniProtKB:P11802')
    validation_results['function'] = {
        'pathways': pathways,
        'go_terms': go_terms
    }
    
    # 4. DRUGGABILITY
    tractability = tu.tools.OpenTargets_get_target_tractability_by_ensemblID(
        ensemblID=ensembl_id
    )
    existing_drugs = tu.tools.OpenTargets_get_associated_drugs_by_target_ensemblID(
        ensemblID=ensembl_id
    )
    validation_results['druggability'] = {
        'tractability': tractability,
        'existing_drugs': existing_drugs
    }
    
    # 5. SAFETY
    safety = tu.tools.OpenTargets_get_target_safety_profile_by_ensemblID(
        ensemblID=ensembl_id
    )
    mouse_models = tu.tools.OpenTargets_get_biological_mouse_models_by_ensemblID(
        ensemblID=ensembl_id
    )
    validation_results['safety'] = {
        'profile': safety,
        'mouse_models': mouse_models
    }
    
    # 6. COMPETITIVE LANDSCAPE
    lit_count = tu.tools.PubMed_search_articles(
        query=f'{gene_symbol} AND drug AND cancer',
        limit=0
    )
    validation_results['competitive'] = {
        'publication_count': lit_count.get('count', 0),
        'maturity': 'mature' if lit_count.get('count', 0) > 1000 else 'emerging'
    }
    
    return validation_results

results = validate_target('CDK4')

Validation Report:

# Target Validation Report: CDK4

## Validation Scorecard

| Criterion | Score | Evidence |
|-----------|-------|----------|
| Genetic Evidence | ✅ HIGH | Strong cancer associations |
| Expression | ✅ HIGH | Overexpressed in multiple cancers |
| Functional Role | ✅ HIGH | Cell cycle regulator |
| Druggability | ✅ HIGH | Approved inhibitors |
| Safety | ⚠️ MEDIUM | On-target effects expected |
| Competitive | 🔴 HIGH | Mature, crowded space |

## Existing Drugs
- Palbociclib (Ibrance)
- Ribociclib (Kisqali)
- Abemaciclib (Verzenio)

## Recommendation
CDK4 is a **validated target** with approved drugs. 
New entrants would need differentiation (selectivity, CNS penetration, etc.)

---

Example 5: Finding Drug Targets for a Disease

Query: "What are the best drug targets for Alzheimer's disease?"

Disease-to-Target Discovery

def find_targets_for_disease(disease_name):
    tu = ToolUniverse(use_cache=True)
    tu.load_tools()
    
    # 1. Get disease ID
    disease_search = tu.tools.OpenTargets_get_disease_ids_by_name(
        diseaseName=disease_name
    )
    efo_id = disease_search.get('id')  # e.g., EFO_0000249 for Alzheimer's
    
    # 2. Get associated targets
    targets = tu.tools.OpenTargets_get_associated_targets_by_disease_efoId(
        efoId=efo_id
    )
    
    # 3. For top targets, assess druggability
    top_targets = targets.get('data', [])[:10]
    
    target_assessments = []
    for target in top_targets:
        ensembl_id = target.get('target_id')
        symbol = target.get('gene_symbol')
        
        # Druggability
        tract = tu.tools.OpenTargets_get_target_tractability_by_ensemblID(
            ensemblID=ensembl_id
        )
        
        # Existing drugs
        drugs = tu.tools.OpenTargets_get_associated_drugs_by_target_ensemblID(
            ensemblID=ensembl_id
        )
        
        # Safety
        safety = tu.tools.OpenTargets_get_target_safety_profile_by_ensemblID(
            ensemblID=ensembl_id
        )
        
        target_assessments.append({
            'symbol': symbol,
            'ensembl_id': ensembl_id,
            'disease_score': target.get('score'),
            'tractability': tract,
            'drug_count': len(drugs.get('data', [])),
            'safety_flags': len(safety.get('data', []))
        })
    
    return target_assessments

alzheimer_targets = find_targets_for_disease("Alzheimer's disease")

Output:

# Top Drug Targets for Alzheimer's Disease

| Rank | Target | Score | Druggability | Drugs | Safety |
|------|--------|-------|--------------|-------|--------|
| 1 | APP | 0.95 | MEDIUM | 0 | ⚠️ |
| 2 | PSEN1 | 0.92 | LOW | 0 | ⚠️ |
| 3 | APOE | 0.88 | LOW | 0 | ✅ |
| 4 | MAPT | 0.85 | MEDIUM | 2 | ⚠️ |
| 5 | BACE1 | 0.82 | HIGH | 0* | ⚠️ |
| 6 | GSK3B | 0.75 | HIGH | 5 | ⚠️ |
| 7 | ACE | 0.70 | HIGH | 10+ | ✅ |
| 8 | TREM2 | 0.68 | MEDIUM | 2 | ✅ |

*BACE1 inhibitors failed in trials

## Recommendations
1. **TREM2** - Emerging target, favorable safety, antibody approaches
2. **GSK3B** - Druggable, existing tool compounds
3. **ACE** - Repurposing opportunity (existing drugs)

---

Quick Reference: Common Multi-Step Patterns

Pattern: Gene Symbol → Full Profile

# 1. Symbol → UniProt
search = tu.tools.UniProt_search(query=f'gene:{symbol} AND organism_id:9606', limit=1)
uniprot = search['results'][0]['primaryAccession']

# 2. UniProt → Ensembl
mapping = tu.tools.UniProt_id_mapping(ids=[uniprot], from_db='UniProtKB_AC-ID', to_db='Ensembl')
ensembl = mapping['results'][0]['to']

# 3. Get all info
entry = tu.tools.UniProt_get_entry_by_accession(accession=uniprot)
tractability = tu.tools.OpenTargets_get_target_tractability_by_ensemblID(ensemblID=ensembl)

Pattern: PDB → Ligand Analysis

# 1. Get PDB info
pdb_info = tu.tools.get_protein_metadata_by_pdb_id(pdb_id='1M17')

# 2. Get ligands
ligands = pdb_info.get('rcsb_binding_affinity', [])

# 3. For each ligand, get ChEMBL data
for lig in ligands:
    comp_id = lig.get('comp_id')
    smiles = tu.tools.get_ligand_smiles_by_chem_comp_id(chem_comp_id=comp_id)
    # Search ChEMBL for similar molecules

Pattern: Disease → Target → Drug

# 1. Disease → EFO ID
efo = tu.tools.OpenTargets_get_disease_ids_by_name(diseaseName='lung cancer')

# 2. EFO → Targets
targets = tu.tools.OpenTargets_get_associated_targets_by_disease_efoId(efoId=efo['id'])

# 3. Target → Drugs
for target in targets['data'][:5]:
    drugs = tu.tools.OpenTargets_get_associated_drugs_by_target_ensemblID(
        ensemblID=target['target_id']
    )

Target Intelligence Implementation Details

Detailed code implementations for each research path. See SKILL.md for the workflow overview.

Identifier Resolution

CRITICAL: Resolve ALL identifiers before any research path.

def resolve_target_ids(tu, query):
    """
    Resolve target query to ALL needed identifiers.
    Returns dict with: query, uniprot, ensembl, ensembl_version, symbol,
    entrez, chembl_target, hgnc
    """
    ids = {
        'query': query,
        'uniprot': None,
        'ensembl': None,
        'ensembl_versioned': None,  # For GTEx
        'symbol': None,
        'entrez': None,
        'chembl_target': None,
        'hgnc': None,
        'full_name': None,
        'synonyms': []
    }

    # [Resolution logic based on input type]
    # ... (see current implementation)

    # CRITICAL: Get versioned Ensembl ID for GTEx
    if ids['ensembl']:
        gene_info = tu.tools.ensembl_lookup_gene(id=ids['ensembl'], species="human")
        if gene_info and gene_info.get('version'):
            ids['ensembl_versioned'] = f"{ids['ensembl']}.{gene_info['version']}"

        # Also get synonyms for literature collision detection
        ids['full_name'] = gene_info.get('description', '').split(' [')[0]

    # Get UniProt alternative names for synonyms
    if ids['uniprot']:
        alt_names = tu.tools.UniProt_get_alternative_names_by_accession(accession=ids['uniprot'])
        if alt_names:
            ids['synonyms'].extend(alt_names)

    return ids

GPCR Target Detection

~35% of approved drugs target GPCRs. After identifier resolution, check if target is a GPCR:

def check_gpcr_target(tu, ids):
    """
    Check if target is a GPCR and retrieve specialized data.
    Call after identifier resolution.
    """
    symbol = ids.get('symbol', '')

    # Build GPCRdb entry name
    entry_name = f"{symbol.lower()}_human"

    gpcr_info = tu.tools.GPCRdb_get_protein(
        operation="get_protein",
        protein=entry_name
    )

    if gpcr_info.get('status') == 'success':
        # Target is a GPCR - get specialized data
        structures = tu.tools.GPCRdb_get_structures(
            operation="get_structures",
            protein=entry_name
        )
        ligands = tu.tools.GPCRdb_get_ligands(
            operation="get_ligands",
            protein=entry_name
        )
        mutations = tu.tools.GPCRdb_get_mutations(
            operation="get_mutations",
            protein=entry_name
        )

        return {
            'is_gpcr': True,
            'gpcr_family': gpcr_info['data'].get('family'),
            'gpcr_class': gpcr_info['data'].get('receptor_class'),
            'structures': structures.get('data', {}).get('structures', []),
            'ligands': ligands.get('data', {}).get('ligands', []),
            'mutations': mutations.get('data', {}).get('mutations', []),
            'ballesteros_numbering': True
        }

    return {'is_gpcr': False}

GPCRdb Report Section (add to Section 2 for GPCR targets):

### 2.x GPCR-Specific Data (GPCRdb)

**Receptor Class**: Class A (Rhodopsin-like)
**GPCR Family**: Adrenoceptors

**Structures by State**:
| PDB ID | State | Resolution | Ligand | Year |
|--------|-------|------------|--------|------|
| 3SN6 | Active | 3.2A | Agonist (BI-167107) | 2011 |
| 2RH1 | Inactive | 2.4A | Antagonist (carazolol) | 2007 |

**Known Ligands**: 45 agonists, 32 antagonists, 8 allosteric modulators
**Key Binding Site Residues** (Ballesteros-Weinstein): 3.32, 5.42, 6.48, 7.39

Collision Detection for Literature Search

Before literature search, detect naming collisions:

def detect_collisions(tu, symbol, full_name):
    """
    Detect if gene symbol has naming collisions in literature.
    Returns negative filter terms if collisions found.
    """
    results = tu.tools.PubMed_search_articles(
        query=f'"{symbol}"[Title]',
        limit=20
    )

    off_topic_terms = []
    for paper in results.get('articles', []):
        title = paper.get('title', '').lower()
        bio_terms = ['protein', 'gene', 'cell', 'expression', 'mutation', 'kinase', 'receptor']
        if not any(term in title for term in bio_terms):
            pass  # Extract potential collision terms

    collision_filter = ""
    if off_topic_terms:
        collision_filter = " NOT " + " NOT ".join(off_topic_terms)

    return collision_filter

PATH 0: Open Targets Foundation

def path_0_open_targets(tu, ids):
    """
    Open Targets foundation data - fills gaps for sections 5, 6, 8, 9, 10, 11.
    ALWAYS run this first.
    """
    ensembl_id = ids['ensembl']
    if not ensembl_id:
        return {'status': 'skipped', 'reason': 'No Ensembl ID'}

    results = {}

    # 1. Diseases & Phenotypes (Section 8)
    diseases = tu.tools.OpenTargets_get_diseases_phenotypes_by_target_ensemblId(
        ensemblId=ensembl_id
    )
    results['diseases'] = diseases if diseases else {'note': 'No disease associations returned'}

    # 2. Tractability (Section 9)
    tractability = tu.tools.OpenTargets_get_target_tractability_by_ensemblID(
        ensemblId=ensembl_id
    )
    results['tractability'] = tractability if tractability else {'note': 'No tractability data returned'}

    # 3. Safety Profile (Section 10)
    safety = tu.tools.OpenTargets_get_target_safety_profile_by_ensemblID(
        ensemblId=ensembl_id
    )
    results['safety'] = safety if safety else {'note': 'No safety liabilities identified'}

    # 4. Interactions (Section 6)
    interactions = tu.tools.OpenTargets_get_target_interactions_by_ensemblID(
        ensemblId=ensembl_id
    )
    results['interactions'] = interactions if interactions else {'note': 'No interactions returned'}

    # 5. GO Annotations (Section 5)
    go_terms = tu.tools.OpenTargets_get_target_gene_ontology_by_ensemblID(
        ensemblId=ensembl_id
    )
    results['go_terms'] = go_terms if go_terms else {'note': 'No GO annotations returned'}

    # 6. Publications (Section 11)
    publications = tu.tools.OpenTargets_get_publications_by_target_ensemblID(
        ensemblId=ensembl_id
    )
    results['publications'] = publications if publications else {'note': 'No publications returned'}

    # 7. Mouse Models (Section 8/10)
    mouse_models = tu.tools.OpenTargets_get_biological_mouse_models_by_ensemblID(
        ensemblId=ensembl_id
    )
    results['mouse_models'] = mouse_models if mouse_models else {'note': 'No mouse model data returned'}

    # 8. Chemical Probes (Section 9)
    probes = tu.tools.OpenTargets_get_chemical_probes_by_target_ensemblID(
        ensemblId=ensembl_id
    )
    results['chemical_probes'] = probes if probes else {'note': 'No chemical probes available'}

    # 9. Associated Drugs (Section 9)
    drugs = tu.tools.OpenTargets_get_associated_drugs_by_target_ensemblID(
        ensemblId=ensembl_id
    )
    results['drugs'] = drugs if drugs else {'note': 'No approved/trial drugs found'}

    return results

Negative Results Are Data

Always document when a query returns empty:

### 9.3 Chemical Probes

**Status**: No validated chemical probes available for this target.
*Source: OpenTargets_get_chemical_probes_by_target_ensemblID returned empty*

**Implication**: Tool compound development would be needed for chemical biology studies.

PATH 2: Structure & Domains (3-Step Chain)

Do NOT rely solely on PDB text search. Use this chain:

def path_structure_robust(tu, ids):
    """Robust structure search using 3-step chain."""
    structures = {'pdb': [], 'alphafold': None, 'domains': [], 'method_notes': []}

    # STEP 1: UniProt PDB Cross-References (most reliable)
    if ids['uniprot']:
        entry = tu.tools.UniProt_get_entry_by_accession(accession=ids['uniprot'])
        pdb_xrefs = [x for x in entry.get('uniProtKBCrossReferences', [])
                    if x.get('database') == 'PDB']
        for xref in pdb_xrefs:
            pdb_id = xref.get('id')
            pdb_info = tu.tools.get_protein_metadata_by_pdb_id(pdb_id=pdb_id)
            if pdb_info:
                structures['pdb'].append(pdb_info)
        structures['method_notes'].append(f"Step 1: {len(pdb_xrefs)} PDB cross-refs from UniProt")

    # STEP 2: Sequence-based PDB Search (catches missing annotations)
    if ids['uniprot'] and len(structures['pdb']) < 5:
        sequence = tu.tools.UniProt_get_sequence_by_accession(accession=ids['uniprot'])
        if sequence and len(sequence) < 1000:
            similar = tu.tools.PDB_search_similar_structures(
                sequence=sequence[:500],
                identity_cutoff=0.7
            )
            if similar:
                for hit in similar[:10]:
                    if hit['pdb_id'] not in [s.get('pdb_id') for s in structures['pdb']]:
                        structures['pdb'].append(hit)
        structures['method_notes'].append(f"Step 2: Sequence search (identity >= 70%)")

    # STEP 3: Domain-based Search (for multi-domain proteins)
    if ids['uniprot']:
        domains = tu.tools.InterPro_get_protein_domains(uniprot_accession=ids['uniprot'])
        structures['domains'] = domains if domains else []

    # AlphaFold (always check)
    alphafold = tu.tools.alphafold_get_prediction(uniprot_accession=ids['uniprot'])
    structures['alphafold'] = alphafold if alphafold else {'note': 'No AlphaFold prediction'}

    # Document limitations
    if not structures['pdb']:
        structures['limitation'] = "No direct PDB hit does NOT mean no structure exists. Check: (1) structures under different UniProt entries, (2) homolog structures, (3) domain-only structures."

    return structures

PATH 5: Expression Profile (GTEx Versioned ID Fallback)

def path_expression(tu, ids):
    """Expression data with GTEx versioned ID fallback."""
    results = {'gtex': None, 'hpa': None, 'failed_tools': []}

    ensembl_id = ids['ensembl']
    versioned_id = ids.get('ensembl_versioned')

    # Try unversioned first
    gtex_result = tu.tools.GTEx_get_median_gene_expression(
        gencode_id=ensembl_id,
        operation="median"
    )

    # Fallback to versioned if empty
    if not gtex_result or gtex_result.get('data') == []:
        if versioned_id:
            gtex_result = tu.tools.GTEx_get_median_gene_expression(
                gencode_id=versioned_id,
                operation="median"
            )
            if gtex_result and gtex_result.get('data'):
                results['gtex'] = gtex_result
                results['gtex_note'] = f"Used versioned ID: {versioned_id}"

        if not results.get('gtex'):
            results['failed_tools'].append({
                'tool': 'GTEx_get_median_gene_expression',
                'tried': [ensembl_id, versioned_id],
                'fallback': 'See HPA data below'
            })
    else:
        results['gtex'] = gtex_result

    # HPA (always query as backup)
    hpa_result = tu.tools.HPA_get_rna_expression_by_source(ensembl_id=ensembl_id)
    results['hpa'] = hpa_result if hpa_result else {'note': 'No HPA RNA data'}

    return results

HPA Extended Expression

def get_hpa_comprehensive_expression(tu, gene_symbol):
    """
    Get comprehensive expression data from Human Protein Atlas.
    Provides tissue expression, subcellular localization, cell line comparison, tissue specificity.
    """
    gene_info = tu.tools.HPA_search_genes_by_query(search_query=gene_symbol)
    if not gene_info:
        return {'error': f'Gene {gene_symbol} not found in HPA'}

    tissue_search = tu.tools.HPA_generic_search(
        search_query=gene_symbol,
        columns="g,gs,rnat,rnatsm,scml,scal",
        format="json"
    )

    cell_lines = ['a549', 'mcf7', 'hela', 'hepg2', 'pc3']
    cell_line_expression = {}
    for cell_line in cell_lines:
        try:
            expr = tu.tools.HPA_get_comparative_expression_by_gene_and_cellline(
                gene_name=gene_symbol,
                cell_line=cell_line
            )
            cell_line_expression[cell_line] = expr
        except:
            continue

    return {
        'gene_info': gene_info,
        'tissue_data': tissue_search,
        'cell_line_expression': cell_line_expression,
        'source': 'Human Protein Atlas'
    }

PATH 6: Variants & Disease

DisGeNET Integration

DisGeNET provides curated gene-disease associations with evidence scores. Requires: DISGENET_API_KEY

def get_disgenet_associations(tu, ids):
    """Get gene-disease associations from DisGeNET."""
    symbol = ids.get('symbol')
    if not symbol:
        return {'status': 'skipped', 'reason': 'No gene symbol'}

    gda = tu.tools.DisGeNET_search_gene(
        operation="search_gene",
        gene=symbol,
        limit=50
    )

    if gda.get('status') != 'success':
        return {'status': 'error', 'message': 'DisGeNET query failed'}

    associations = gda.get('data', {}).get('associations', [])
    strong, moderate, weak = [], [], []

    for assoc in associations:
        score = assoc.get('score', 0)
        entry = {
            'disease': assoc.get('disease_name', ''),
            'umls_cui': assoc.get('disease_id', ''),
            'score': score,
            'evidence_index': assoc.get('ei'),
            'dsi': assoc.get('dsi'),
            'dpi': assoc.get('dpi')
        }
        if score >= 0.7:
            strong.append(entry)
        elif score >= 0.4:
            moderate.append(entry)
        else:
            weak.append(entry)

    return {
        'total_associations': len(associations),
        'strong_associations': strong,
        'moderate_associations': moderate,
        'weak_associations': weak[:10],
        'disease_pleiotropy': len(associations)
    }

Evidence Tier Assignment:

• DisGeNET Score >= 0.7 -> Consider T2 evidence (multiple validated sources)
• DisGeNET Score 0.4-0.7 -> Consider T3 evidence
• DisGeNET Score < 0.4 -> T4 evidence only

PATH 7: Druggability & Target Validation

Pharos/TCRD - Target Development Level

def get_pharos_target_info(tu, ids):
    """Get Pharos/TCRD target development level and druggability."""
    gene_symbol = ids.get('symbol')
    uniprot = ids.get('uniprot')

    if gene_symbol:
        result = tu.tools.Pharos_get_target(gene=gene_symbol)
    elif uniprot:
        result = tu.tools.Pharos_get_target(uniprot=uniprot)
    else:
        return {'status': 'error', 'message': 'Need gene symbol or UniProt'}

    if result.get('status') == 'success' and result.get('data'):
        target = result['data']
        return {
            'name': target.get('name'),
            'symbol': target.get('sym'),
            'tdl': target.get('tdl'),
            'family': target.get('fam'),
            'novelty': target.get('novelty'),
            'description': target.get('description'),
            'publications': target.get('publicationCount'),
            'interpretation': interpret_tdl(target.get('tdl'))
        }
    return None

def interpret_tdl(tdl):
    interpretations = {
        'Tclin': 'Approved drug target - highest confidence for druggability',
        'Tchem': 'Small molecule active - good chemical tractability',
        'Tbio': 'Biologically characterized - may require novel modalities',
        'Tdark': 'Understudied - limited data, high novelty potential'
    }
    return interpretations.get(tdl, 'Unknown')

def search_disease_targets(tu, disease_name):
    """Find targets associated with a disease via Pharos."""
    result = tu.tools.Pharos_get_disease_targets(disease=disease_name, top=50)
    if result.get('status') == 'success':
        targets = result['data'].get('targets', [])
        by_tdl = {'Tclin': [], 'Tchem': [], 'Tbio': [], 'Tdark': []}
        for t in targets:
            tdl = t.get('tdl', 'Unknown')
            if tdl in by_tdl:
                by_tdl[tdl].append(t)
        return by_tdl
    return None

DepMap - Target Essentiality

def assess_target_essentiality(tu, ids):
    """
    Is this target essential for cancer cell survival?
    Negative effect scores = gene is essential (cells die upon KO)
    """
    gene_symbol = ids.get('symbol')
    if not gene_symbol:
        return {'status': 'error', 'message': 'Need gene symbol'}

    deps = tu.tools.DepMap_get_gene_dependencies(gene_symbol=gene_symbol)

    if deps.get('status') == 'success':
        return {
            'gene': gene_symbol,
            'data': deps.get('data', {}),
            'interpretation': 'Negative scores indicate gene is essential for cell survival',
            'note': 'Score < -0.5 is strongly essential, < -1.0 is extremely essential'
        }
    return None

InterProScan - Novel Domain Prediction

For uncharacterized proteins, run InterProScan to predict domains and function:

def predict_protein_domains(tu, sequence, title="Query protein"):
    """
    Run InterProScan for de novo domain prediction.
    Use when: Novel/uncharacterized proteins, custom sequences, sparse InterPro annotations.
    """
    result = tu.tools.InterProScan_scan_sequence(
        sequence=sequence,
        title=title,
        go_terms=True,
        pathways=True
    )

    if result.get('status') == 'success':
        data = result.get('data', {})
        if data.get('job_status') == 'RUNNING':
            return {
                'job_id': data.get('job_id'),
                'status': 'running',
                'note': 'Use InterProScan_get_job_results to retrieve when ready'
            }
        return {
            'domains': data.get('domains', []),
            'domain_count': data.get('domain_count', 0),
            'go_annotations': data.get('go_annotations', []),
            'pathways': data.get('pathways', []),
            'sequence_length': data.get('sequence_length')
        }
    return None

BindingDB - Known Ligands & Binding Data

def get_bindingdb_ligands(tu, uniprot_id, affinity_cutoff=10000):
    """
    Get ligands with measured binding affinities from BindingDB.
    Critical for identifying chemical starting points and assessing tractability.
    """
    result = tu.tools.BindingDB_get_ligands_by_uniprot(
        uniprot=uniprot_id,
        affinity_cutoff=affinity_cutoff
    )

    if result:
        ligands = []
        for entry in result:
            ligands.append({
                'smiles': entry.get('smile'),
                'affinity_type': entry.get('affinity_type'),
                'affinity_nM': entry.get('affinity'),
                'monomer_id': entry.get('monomerid'),
                'pmid': entry.get('pmid')
            })
        ligands.sort(key=lambda x: float(x['affinity_nM']) if x['affinity_nM'] else float('inf'))
        return {
            'total_ligands': len(ligands),
            'ligands': ligands[:20],
            'best_affinity': ligands[0]['affinity_nM'] if ligands else None
        }

    return {'total_ligands': 0, 'ligands': [], 'note': 'No ligands found in BindingDB'}

Affinity Interpretation:

PubChem BioAssay - Screening Data

def get_pubchem_assays_for_target(tu, gene_symbol):
    """Get bioassays targeting a gene from PubChem."""
    assays = tu.tools.PubChem_search_assays_by_target_gene(gene_symbol=gene_symbol)

    assay_info = []
    if assays.get('data', {}).get('aids'):
        for aid in assays['data']['aids'][:10]:
            summary = tu.tools.PubChem_get_assay_summary(aid=aid)
            targets = tu.tools.PubChem_get_assay_targets(aid=aid)
            assay_info.append({
                'aid': aid,
                'summary': summary.get('data', {}),
                'targets': targets.get('data', {})
            })

    return {
        'total_assays': len(assays.get('data', {}).get('aids', [])),
        'assay_details': assay_info
    }

PATH 8: Literature (Collision-Aware)

def path_literature_collision_aware(tu, ids):
    """Literature search with collision detection and filtering."""
    symbol = ids['symbol']
    full_name = ids.get('full_name', '')
    uniprot = ids['uniprot']
    synonyms = ids.get('synonyms', [])

    # Step 1: Detect collisions
    collision_filter = detect_collisions(tu, symbol, full_name)

    # Step 2: Build high-precision seed queries
    seed_queries = [
        f'"{symbol}"[Title] AND (protein OR gene OR expression)',
        f'"{full_name}"[Title]' if full_name else None,
        f'"UniProt:{uniprot}"' if uniprot else None,
    ]
    seed_queries = [q for q in seed_queries if q]
    for syn in synonyms[:3]:
        seed_queries.append(f'"{syn}"[Title]')

    # Step 3: Execute seed queries and collect PMIDs
    seed_pmids = set()
    for query in seed_queries:
        if collision_filter:
            query = f"({query}){collision_filter}"
        results = tu.tools.PubMed_search_articles(query=query, limit=30)
        for article in results.get('articles', []):
            seed_pmids.add(article.get('pmid'))

    # Step 4: Expand via citation network (for sparse targets)
    if len(seed_pmids) < 30:
        expanded_pmids = set()
        for pmid in list(seed_pmids)[:10]:
            related = tu.tools.PubMed_get_related(pmid=pmid, limit=20)
            for r in related.get('articles', []):
                expanded_pmids.add(r.get('pmid'))
            citing = tu.tools.EuropePMC_get_citations(pmid=pmid, limit=20)
            for c in citing.get('citations', []):
                expanded_pmids.add(c.get('pmid'))
        seed_pmids.update(expanded_pmids)

    # Step 5: Classify papers by evidence tier
    papers_by_tier = {'T1': [], 'T2': [], 'T3': [], 'T4': []}

    return {
        'total_papers': len(seed_pmids),
        'collision_filter_applied': collision_filter if collision_filter else 'None needed',
        'seed_queries': seed_queries,
        'papers_by_tier': papers_by_tier
    }

Retry Logic & Fallback Chains

Retry Policy

def call_with_retry(tu, tool_name, params, max_retries=3):
    """Call tool with retry logic."""
    for attempt in range(max_retries):
        try:
            result = getattr(tu.tools, tool_name)(**params)
            if result and not result.get('error'):
                return result
        except Exception as e:
            if attempt < max_retries - 1:
                time.sleep(2 ** attempt)
            else:
                return {'error': str(e), 'tool': tool_name, 'attempts': max_retries}
    return None

Fallback Chains

Failure Surfacing Rule

NEVER silently skip failed tools. Always document:

### 7.1 Tissue Expression

**GTEx Data**: Unavailable (API timeout after 3 attempts)
**Fallback Data (HPA)**:
| Tissue | Expression Level | Specificity |
|--------|-----------------|-------------|
| Liver | High | Enhanced |
| Kidney | Medium | - |

*Note: For complete GTEx data, query directly at gtexportal.org*

Target Intelligence Tool Reference

Complete reference of 225+ ToolUniverse tools for target research, organized by category.

1. Core Protein Information (UniProt)

EBI Proteins API

2. Gene Information

MyGene (BioThings)

Ensembl

Other Gene Resources

3. Drug-Target Interactions

DGIdb

ChEMBL

DrugBank & GtoPdb

STITCH

GPCRdb (NEW - for GPCR Targets)

~35% of approved drugs target GPCRs. GPCRdb provides specialized data for G protein-coupled receptors.

Entry name format: {gene_lower}_human (e.g., adrb2_human, drd2_human)

Key advantages:

• Active vs. inactive state structures
• Ballesteros-Weinstein residue numbering
• Curated ligand binding data
• Experimental mutation effects

Pharos/TCRD (NEW - Target Development Level)

NIH's Illuminating the Druggable Genome (IDG) portal provides TDL classification.

TDL Classification:

Example:

result = tu.tools.Pharos_get_target(gene="EGFR")
# Returns: tdl="Tclin", fam="Kinase", novelty=0.2, publicationCount=45000

DepMap (NEW - Target Essentiality)

CRISPR knockout essentiality data from cancer cell lines.

Effect Score Interpretation:

Example:

deps = tu.tools.DepMap_get_gene_dependencies(gene_symbol="KRAS")
# Returns: Gene info, note about essentiality

cells = tu.tools.DepMap_get_cell_lines(cancer_type="Lung Cancer", page_size=10)
# Returns: Cell line names, cancer types, MSI status

InterProScan (NEW - Domain Prediction)

De novo domain/family prediction for novel sequences.

When to use: Novel proteins, Tdark targets, custom sequences.

Example:

# Submit sequence for analysis
result = tu.tools.InterProScan_scan_sequence(
    sequence="MVLSPADKTNVKAAWGKVGAHAGEYGAEALERMFLSFPTTKTYFPHFDLSH...",
    go_terms=True,
    pathways=True
)
# Returns: job_id (if running) or domains/GO/pathways (if complete)

# Check job if still running
status = tu.tools.InterProScan_get_job_status(job_id="iprscan5-xxx")
results = tu.tools.InterProScan_get_job_results(job_id="iprscan5-xxx")

BindingDB (NEW - Ligand Binding Data)

Experimental binding affinity data (Ki, IC50, Kd) for target-ligand pairs.

Example:

# Get ligands for EGFR
ligands = tu.tools.BindingDB_get_ligands_by_uniprot(
    uniprot="P00533",
    affinity_cutoff=100  # Only potent ligands <100 nM
)
# Returns: SMILES, affinity_type (Ki/IC50/Kd), affinity value, PMID

# Find targets for a compound
targets = tu.tools.BindingDB_get_targets_by_compound(
    smiles="CC(=O)Nc1ccc(cc1)O",
    similarity_cutoff=0.85
)
# Returns: proteins with similar compound activities

Affinity Interpretation:

Human Protein Atlas (NEW - Expression)

Protein and RNA expression across tissues and cell lines.

Example:

# Search gene
gene = tu.tools.HPA_search_genes_by_query(search_query="EGFR")
# Returns: Gene name, Ensembl ID, synonyms

# Compare cancer cell line vs normal tissue
expr = tu.tools.HPA_get_comparative_expression_by_gene_and_cellline(
    gene_name="EGFR",
    cell_line="a549"  # Lung cancer
)
# Returns: expression comparison

Supported Cell Lines: a549, mcf7, hela, hepg2, pc3, jurkat, rh30, siha, u251, ishikawa

PubChem BioAssay (NEW - Screening Data)

HTS screening data and dose-response curves.

Example:

# Find assays for target
assays = tu.tools.PubChem_search_assays_by_target_gene(gene_symbol="EGFR")
# Returns: list of AIDs

# Get assay summary
summary = tu.tools.PubChem_get_assay_summary(aid=504526)
# Returns: active/inactive counts, target info

# Get active compounds
actives = tu.tools.PubChem_get_assay_active_compounds(aid=504526)
# Returns: CIDs of active compounds

4. Open Targets Platform

Target-Centric

Disease-Target Evidence

5. Protein Structure

RCSB PDB

PDBe

AlphaFold

EMDB

6. Protein-Protein Interactions

STRING

IntAct

Other PPI Sources

7. Functional Annotations

Gene Ontology

InterPro & Pfam

Gene Set Enrichment

8. Pathways

Reactome

KEGG

WikiPathways

Pathway Commons

9. Gene Expression

GTEx

Human Protein Atlas (HPA)

Single-Cell

10. Variants & Mutations

ClinVar

dbSNP

gnomAD

CIViC

Other Variant Sources

11. Literature

PubMed

Europe PMC

Other Literature

12. Pharmacogenomics

PharmGKB

13. Disease Associations

DisGeNET (NEW - Gene-Disease Associations)

DisGeNET integrates gene-disease associations from curated repositories, GWAS catalogs, animal models, and literature. Requires: DISGENET_API_KEY

Key metrics:

• GDA Score: 0-1 confidence score for gene-disease association
• Evidence Index: Number and diversity of sources
• Disease Specificity Index: How specific is gene to this disease
• Disease Pleiotropy Index: How many diseases gene is linked to

Interpretation:

• Score ≥0.7: Strong association (consider T2 evidence)
• Score 0.4-0.7: Moderate association
• Score <0.4: Weak/limited evidence

14. ID Conversion & Cross-References

Common ID Mapping Combinations

Target Intelligence Report Format Specification

This document defines the comprehensive report format for target intelligence reports. ALL sections are REQUIRED unless marked optional.

Critical Requirements

1. Report File Creation

• Create the report file FIRST before any data collection
• File name: [TARGET]_target_report.md (e.g., EGFR_target_report.md)
• Initialize all 14 section headers with [Researching...] placeholders
• Update sections progressively as data is retrieved

2. Citation Requirements (MANDATORY)

Every piece of data MUST cite its source. Include:

• The database name (UniProt, PDB, ChEMBL, etc.)
• The tool used (UniProt_get_entry_by_accession, etc.)
• The query parameters (accession, gene symbol, etc.)

3. Source Block Format

At the end of each major section, add a source block:

---
**Sources:**
- [Database]: `tool_name` (query_parameter)
- [Database]: `tool_name` (query_parameter)
---

Report Template

# Target Intelligence Report: [FULL PROTEIN NAME]

**Generated**: [Date] | **Query**: [Original query] | **Completeness**: [X/8 paths successful]

---

## 1. Executive Summary

[2-3 sentence overview of the target covering: what it is, its primary function, druggability status, and key clinical relevance]

**Bottom Line**: [One sentence: Is this a good drug target? Why/why not?]

---

## 2. Target Identifiers

| Identifier Type | Value | Database |
|-----------------|-------|----------|
| Gene Symbol | [SYMBOL] | HGNC |
| UniProt Accession | [P#####] | UniProtKB |
| Ensembl Gene ID | [ENSG###] | Ensembl |
| Entrez Gene ID | [#####] | NCBI Gene |
| ChEMBL Target ID | [CHEMBL###] | ChEMBL |
| HGNC ID | [HGNC:####] | HGNC |

**Aliases**: [List all known aliases/synonyms]

---

## 3. Basic Information

### 3.1 Protein Description
- **Recommended Name**: [Full protein name]
- **Alternative Names**: [List]
- **Gene Name**: [Symbol] ([Full gene name])
- **Organism**: [Species] (Taxonomy ID: [####])
- **Protein Length**: [###] amino acids
- **Molecular Weight**: [###] kDa
- **Isoforms**: [Number] known isoforms

### 3.2 Protein Function
[Detailed description of protein function - at least 3-4 sentences covering:
- Primary molecular function
- Biological process involvement
- Cellular role
- Signaling pathway context]

### 3.3 Subcellular Localization
- **Primary Location**: [e.g., Plasma membrane]
- **Additional Locations**: [List]
- **Topology**: [e.g., Single-pass type I membrane protein]

---

## 4. Structural Biology

### 4.1 Experimental Structures (PDB)
| PDB ID | Resolution | Method | Ligand | Description |
|--------|------------|--------|--------|-------------|
| [####] | [#.#Å] | [X-ray/Cryo-EM/NMR] | [Ligand or Apo] | [Brief description] |
[List top 5-10 most relevant structures]

**Total PDB Entries**: [###]
**Best Resolution**: [#.#Å] ([PDB ID])
**Structure Coverage**: [Complete/Partial - which domains?]

### 4.2 AlphaFold Prediction
- **Available**: [Yes/No]
- **Confidence**: [High/Medium/Low - pLDDT scores]
- **Model URL**: [AlphaFold DB link]

### 4.3 Domain Architecture
| Domain | Position | InterPro ID | Description |
|--------|----------|-------------|-------------|
| [Domain name] | [Start-End] | [IPR######] | [Function] |
[List all domains]

### 4.4 Key Structural Features
- **Active Sites**: [List with positions]
- **Binding Sites**: [List - substrate, cofactor, drug binding]
- **PTM Sites**: [Phosphorylation, glycosylation, etc. with positions]
- **Disulfide Bonds**: [List]

### 4.5 Structural Druggability Assessment
- **Binding Pockets**: [Identified pockets suitable for small molecules]
- **Allosteric Sites**: [Known or predicted]
- **Antibody Epitopes**: [Surface accessibility for biologics]

---

## 5. Function & Pathways

### 5.1 Gene Ontology Annotations

**Molecular Function (MF)**:
| GO Term | GO ID | Evidence |
|---------|-------|----------|
| [Term] | [GO:#######] | [IDA/IEA/etc.] |
[List top 5-10]

**Biological Process (BP)**:
| GO Term | GO ID | Evidence |
|---------|-------|----------|
| [Term] | [GO:#######] | [IDA/IEA/etc.] |
[List top 5-10]

**Cellular Component (CC)**:
| GO Term | GO ID | Evidence |
|---------|-------|----------|
| [Term] | [GO:#######] | [IDA/IEA/etc.] |
[List top 5]

### 5.2 Pathway Involvement
| Pathway | Database | Pathway ID |
|---------|----------|------------|
| [Pathway name] | [Reactome/KEGG/WikiPathways] | [ID] |
[List top 10 pathways]

### 5.3 Functional Summary
[Paragraph describing the target's role in cellular signaling, disease mechanisms, and biological importance]

---

## 6. Protein-Protein Interactions

### 6.1 Interaction Network Summary
- **Total Interactors (STRING, score >0.7)**: [###]
- **Experimentally Validated (IntAct)**: [###]
- **Complex Membership**: [List complexes]

### 6.2 Top Interacting Partners
| Partner | Score | Interaction Type | Evidence | Biological Context |
|---------|-------|------------------|----------|-------------------|
| [Gene] | [0.###] | [Physical/Functional] | [Experimental/Predicted] | [Context] |
[List top 15-20 interactors]

### 6.3 Protein Complexes
| Complex Name | Members | Function |
|--------------|---------|----------|
| [Complex] | [List] | [Function] |

### 6.4 Interaction Network Implications
[Paragraph on network topology, hub status, and implications for drugging]

---

## 7. Expression Profile

### 7.1 Tissue Expression (GTEx/HPA)
| Tissue | Expression Level (TPM) | Specificity |
|--------|------------------------|-------------|
| [Tissue] | [###] | [High/Medium/Low] |
[List top 10 expressing tissues]

**Tissue Specificity Score**: [Score] ([Broadly expressed/Tissue-specific/Tissue-enriched])

### 7.2 Cell Type Expression
[Single-cell data if available - top cell types]

### 7.3 Disease-Relevant Expression
| Cancer/Disease | Expression Change | Prognostic Value |
|----------------|-------------------|------------------|
| [Disease] | [Up/Down/Unchanged] | [Favorable/Unfavorable/None] |

### 7.4 Expression-Based Druggability
- **Tumor vs Normal**: [Differential expression ratio]
- **Therapeutic Window**: [Assessment based on expression pattern]

---

## 8. Genetic Variation & Disease

### 8.1 Genetic Constraint Scores
| Metric | Value | Interpretation |
|--------|-------|----------------|
| pLI | [0.##] | [Highly constrained/Tolerant] |
| LOEUF | [0.##] | [Interpretation] |
| Missense Z-score | [#.##] | [Interpretation] |
| pRec | [0.##] | [Interpretation] |

### 8.2 Disease Associations (Open Targets)
| Disease | Association Score | Evidence Types | EFO ID |
|---------|-------------------|----------------|--------|
| [Disease] | [0.##] | [Genetic/Literature/etc.] | [EFO_#######] |
[List top 10 diseases]

### 8.3 Pathogenic Variants (ClinVar)
| Variant | Clinical Significance | Condition | Review Status |
|---------|----------------------|-----------|---------------|
| [p.XXX###YYY] | [Pathogenic/Likely pathogenic] | [Condition] | [Stars] |
[List notable pathogenic variants]

**Total ClinVar Entries**: [###]
**Pathogenic/Likely Pathogenic**: [###]

### 8.4 Cancer Mutations (COSMIC/cBioPortal)
| Mutation | Frequency | Cancer Types | Functional Impact |
|----------|-----------|--------------|-------------------|
| [Mutation] | [#%] | [Cancers] | [Activating/Inactivating/Unknown] |
[List recurrent cancer mutations]

### 8.5 Genetic Evidence Summary
[Paragraph summarizing genetic validation of the target]

---

## 9. Druggability & Pharmacology

### 9.1 Tractability Assessment (Open Targets)
| Modality | Tractability | Bucket | Evidence |
|----------|--------------|--------|----------|
| Small Molecule | [✅/⚠️/❌] | [1-10] | [Clinical/Predicted] |
| Antibody | [✅/⚠️/❌] | [1-10] | [Clinical/Predicted] |
| PROTAC | [✅/⚠️/❌] | [1-10] | [Structural feasibility] |
| Other Modalities | [✅/⚠️/❌] | [1-10] | [Evidence] |

### 9.2 Approved Drugs
| Drug Name | Brand Name | Mechanism | Indication | Approval Year |
|-----------|------------|-----------|------------|---------------|
| [Drug] | [Brand] | [Inhibitor/Agonist/etc.] | [Indication] | [Year] |
[List all approved drugs]

### 9.3 Clinical Pipeline
| Drug | Phase | Indication | Trial Count | Status |
|------|-------|------------|-------------|--------|
| [Drug] | [Phase I/II/III] | [Indication] | [###] | [Active/Completed] |
[List drugs in clinical development]

**Total Clinical Trials**: [###]
**Active Trials**: [###]

### 9.4 Bioactivity Data (ChEMBL)
- **Total Bioactivity Records**: [###]
- **Compounds Tested**: [###]
- **Most Potent Compound**: [Name] (IC50/Ki: [###] nM)

| Compound | Activity Type | Value | Target |
|----------|---------------|-------|--------|
| [Compound] | [IC50/Ki/Kd] | [###] nM | [Target form] |
[List top 5 most potent compounds]

### 9.5 Chemical Probes
| Probe | Selectivity | Use | Source |
|-------|-------------|-----|--------|
| [Probe] | [Selective/Broad] | [Recommended use] | [SGC/etc.] |

### 9.6 Drug Resistance
[Known resistance mechanisms, mutations, or mechanisms if applicable]

---

## 10. Safety Profile

### 10.1 Target Safety Liabilities (Open Targets)
| Safety Concern | Evidence | Severity | Organ System |
|----------------|----------|----------|--------------|
| [Concern] | [Animal/Human/Both] | [High/Medium/Low] | [System] |

### 10.2 Mouse Knockout Phenotypes
| Phenotype | Zygosity | Viability | Source |
|-----------|----------|-----------|--------|
| [Phenotype] | [Homo/Hetero] | [Viable/Lethal] | [IMPC/MGI] |

### 10.3 Known Drug Adverse Events
| Adverse Event | Frequency | Drug Class | Mechanism |
|---------------|-----------|------------|-----------|
| [Event] | [Common/Uncommon/Rare] | [Class] | [On-target/Off-target] |

### 10.4 Safety Summary
[Paragraph summarizing safety considerations for targeting this protein]

---

## 11. Literature & Research Landscape

### 11.1 Publication Metrics
| Metric | Value |
|--------|-------|
| Total Publications | [###,###] |
| Publications (Last 5 Years) | [###,###] |
| Publications (Last Year) | [###,###] |
| Drug-Related Publications | [###,###] |
| Clinical Publications | [###,###] |

### 11.2 Research Trend
- **Trend**: [Increasing/Stable/Declining]
- **Peak Year**: [Year]
- **Current Activity**: [High/Medium/Low]

### 11.3 Key Research Areas
[List current hot topics in research for this target]

### 11.4 Notable Recent Publications
| PMID | Title | Year | Key Finding |
|------|-------|------|-------------|
| [PMID] | [Title] | [Year] | [Finding] |
[List 3-5 important recent papers]

---

## 12. Competitive Landscape

### 12.1 Market Status
- **First-in-Class Approved**: [Yes/No - Drug name if yes]
- **Best-in-Class Status**: [Assessment]
- **Patent Landscape**: [Crowded/Moderate/Open]

### 12.2 Differentiation Opportunities
[List potential differentiation strategies for new entrants]

---

## 13. Summary & Recommendations

### 13.1 Target Validation Scorecard
| Criterion | Score (1-5) | Evidence Level |
|-----------|-------------|----------------|
| Genetic Evidence | [#] | [Strong/Moderate/Weak] |
| Expression Relevance | [#] | [Strong/Moderate/Weak] |
| Functional Understanding | [#] | [Strong/Moderate/Weak] |
| Druggability | [#] | [Strong/Moderate/Weak] |
| Safety Profile | [#] | [Favorable/Caution/Concern] |
| Competitive Position | [#] | [Open/Moderate/Crowded] |

**Overall Target Score**: [##/30]

### 13.2 Key Strengths
1. [Strength 1]
2. [Strength 2]
3. [Strength 3]

### 13.3 Key Challenges/Risks
1. [Challenge 1]
2. [Challenge 2]
3. [Challenge 3]

### 13.4 Recommendations
| Priority | Category | Recommendation |
|----------|----------|----------------|
| 🔴 HIGH | [Category] | [Recommendation] |
| 🟡 MEDIUM | [Category] | [Recommendation] |
| 🟢 LOW | [Category] | [Recommendation] |
| ℹ️ INFO | [Category] | [Recommendation] |

### 13.5 Next Steps
[Suggested follow-up analyses or experiments]

---

## 14. Data Sources & Methodology

### 14.1 Databases Queried
| Database | Section(s) | Queries | Status |
|----------|------------|---------|--------|
| UniProtKB | 2, 3, 4, 8 | [accession] | ✅ Success |
| RCSB PDB | 4 | [PDB IDs queried] | ✅ Success |
| AlphaFold DB | 4 | [accession] | ✅ Success |
| InterPro | 4 | [accession] | ✅ Success |
| Gene Ontology | 5 | [gene_id] | ✅ Success |
| Reactome | 5 | [accession] | ✅ Success |
| KEGG | 5 | [gene_id] | ✅ Success |
| STRING | 6 | [protein_ids] | ✅ Success |
| IntAct | 6 | [accession] | ✅ Success |
| GTEx | 7 | [gencode_id] | ✅ Success |
| Human Protein Atlas | 7 | [ensembl_id] | ✅ Success |
| gnomAD | 8 | [gene_symbol] | ✅ Success |
| ClinVar | 8 | [gene] | ✅ Success |
| Open Targets | 8, 9, 10 | [ensembl_id] | ✅ Success |
| ChEMBL | 9 | [target_chembl_id] | ✅ Success |
| DGIdb | 9 | [genes] | ✅ Success |
| PubMed | 11 | [query] | ✅ Success |

### 14.2 Tools Used by Section
| Section | Tools Used |
|---------|------------|
| 2. Identifiers | `UniProt_search`, `UniProt_id_mapping`, `MyGene_get_gene_annotation` |
| 3. Basic Info | `UniProt_get_entry_by_accession`, `UniProt_get_function_by_accession`, `UniProt_get_subcellular_location_by_accession` |
| 4. Structure | `get_protein_metadata_by_pdb_id`, `alphafold_get_prediction`, `InterPro_get_protein_domains`, `UniProt_get_ptm_processing_by_accession` |
| 5. Function | `GO_get_annotations_for_gene`, `Reactome_map_uniprot_to_pathways`, `kegg_get_gene_info` |
| 6. Interactions | `STRING_get_protein_interactions`, `intact_get_interactions`, `intact_get_complex_details` |
| 7. Expression | `GTEx_get_median_gene_expression`, `HPA_get_comprehensive_gene_details_by_ensembl_id`, `HPA_get_subcellular_location` |
| 8. Variants | `gnomad_get_gene_constraints`, `ClinVar_search_variants`, `OpenTargets_get_diseases_phenotypes_by_target_ensembl` |
| 9. Druggability | `OpenTargets_get_target_tractability_by_ensemblID`, `OpenTargets_get_associated_drugs_by_target_ensemblID`, `DGIdb_get_gene_druggability`, `ChEMBL_get_target_activities` |
| 10. Safety | `OpenTargets_get_target_safety_profile_by_ensemblID`, `OpenTargets_get_biological_mouse_models_by_ensemblID` |
| 11. Literature | `PubMed_search_articles`, `EuropePMC_search_articles`, `OpenTargets_get_publications_by_target_ensemblID` |

### 14.3 Data Freshness
- **Report Generated**: [YYYY-MM-DD HH:MM UTC]
- **UniProt Release**: [Release number if available]
- **PDB Last Updated**: [Date]
- **GTEx Version**: v8
- **gnomAD Version**: v4.0

### 14.4 Limitations & Data Gaps
[Document any issues encountered:]
- Tools that returned errors or empty results
- Sections with incomplete data
- Known data quality issues
- Databases that were unavailable

Example:
- ⚠️ IntAct returned no experimentally validated interactions
- ⚠️ Some ChEMBL activity data may include non-human orthologs
- ✅ All primary data sources queried successfully

---

## Appendix (Optional)

### A. Full Sequence
[Protein sequence in FASTA format if requested]

### B. Additional Structures
[Extended PDB list if many available]

### C. Complete Interaction List
[Full PPI list if requested]

### D. All Variants
[Complete variant list if requested]

---

Section Completeness Checklist

For each report, verify ALL items:

Required Sections

• [ ] Section 1: Executive Summary with bottom line
• [ ] Section 2: All identifier types (UniProt, Ensembl, Entrez, ChEMBL)
• [ ] Section 3: Basic info with function description (3-4 sentences)
• [ ] Section 4: Structural data (PDB count, domains, AlphaFold)
• [ ] Section 5: GO terms (5-10 per category) and pathways (top 10)
• [ ] Section 6: PPI (15-20 interactors, complexes)
• [ ] Section 7: Expression (top 10 tissues, specificity)
• [ ] Section 8: Variants (constraint scores, top 10 diseases, pathogenic variants)
• [ ] Section 9: Druggability (tractability, all drugs, clinical pipeline)
• [ ] Section 10: Safety (liabilities, mouse KO, adverse events)
• [ ] Section 11: Literature (5 metrics, trend, key papers)
• [ ] Section 12: Competitive landscape
• [ ] Section 13: Scorecard and recommendations
• [ ] Section 14: Data sources and methodology

Data Minimums

• [ ] At least 5 PDB structures listed (if available)
• [ ] All protein domains included
• [ ] Top 10 GO terms per category
• [ ] Top 10 pathways
• [ ] Top 15-20 protein interactors
• [ ] Expression in top 10 tissues
• [ ] All 4 constraint scores (pLI, LOEUF, missense Z, pRec)
• [ ] Top 10 disease associations
• [ ] All approved drugs listed
• [ ] All safety concerns documented
• [ ] 5 publication metrics
• [ ] 3-5 recent key publications
• [ ] Scorecard with all 6 criteria
• [ ] At least 3 prioritized recommendations

Quality Checks

• [ ] Executive summary is 2-3 sentences
• [ ] Function description is 3-4 sentences
• [ ] All tables have data (no empty tables)
• [ ] Paragraphs provide synthesis, not just lists
• [ ] Recommendations are actionable and prioritized
• [ ] Limitations section is honest about data gaps

---

Section-Specific Guidance

Section 1: Executive Summary

Purpose: Give reader the key takeaways in 30 seconds

Include: 1. What the target IS (protein class, function) 2. Clinical relevance (disease associations) 3. Druggability status (has drugs? tractable?) 4. One-line recommendation

Example:

EGFR is a receptor tyrosine kinase that drives cell proliferation and is overexpressed in multiple cancers including NSCLC and glioblastoma. Multiple approved TKIs and monoclonal antibodies validate this as a highly druggable target, though resistance mutations remain a challenge.

>

Bottom Line: Well-validated, druggable target with clinical precedence; new programs should focus on resistance-breaking or novel modalities.

Section 4: Structural Biology

Purpose: Enable structure-based drug design decisions

Must include:

• Total PDB count and best resolution
• Coverage (which domains have structures?)
• AlphaFold availability and confidence
• Complete domain list with positions
• Key binding sites for drug design

Section 9: Druggability

Purpose: Assess feasibility and competitive landscape

Must include:

• Tractability for ALL modalities (SM, Ab, PROTAC, other)
• Complete list of approved drugs
• Clinical pipeline (phase, indication, status)
• ChEMBL bioactivity summary
• Chemical probes if available

Section 13: Recommendations

Purpose: Provide actionable next steps

Requirements:

• Use priority levels (HIGH/MEDIUM/LOW/INFO)
• Each recommendation must be actionable
• Include both opportunities and risks
• Suggest specific follow-up analyses

Example: