Tooluniverse Drug Research

Drug Research Strategy

Comprehensive drug investigation using 50+ ToolUniverse tools across chemical databases, clinical trials, adverse events, pharmacogenomics, and literature.

KEY PRINCIPLES: 1. Report-first approach - Create report file FIRST, then populate progressively 2. Compound disambiguation FIRST - Resolve identifiers before research 3. Citation requirements - Every fact must have inline source attribution 4. Evidence grading - Grade claims by evidence strength (T1-T4) 5. Mandatory completeness - All sections must exist, even if "data unavailable" 6. English-first queries - Always use English drug/compound names in tool calls, even if the user writes in another language. Only try original-language terms as a fallback. Respond in the user's language

---

LOOK UP, DON'T GUESS

When asked about a drug, query ChEMBL/PubChem/DailyMed FIRST. Don't guess at mechanism, targets, or side effects — look them up. 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.

---

Drug Mechanism Reasoning

When investigating a drug's mechanism of action, trace the full causal chain: 1. Target engagement - Which protein(s) does the drug bind, and with what affinity/selectivity? 2. Molecular effect - Does binding inhibit, activate, or modulate the target's function? 3. Pathway consequence - Which signaling or metabolic pathway is altered downstream? 4. Cellular phenotype - What changes occur at the cell level (proliferation, apoptosis, secretion)? 5. Physiological outcome - How does the cellular effect translate to the therapeutic benefit in the patient?

---

Workflow Overview

1. Report-First Approach (MANDATORY)

DO NOT show the search process or tool outputs to the user. Instead:

1. Create the report file FIRST - [DRUG]_drug_report.md with all 11 section headers and [Researching...] placeholders. See REPORT_TEMPLATE.md for the full template. 2. Progressively update the report - Replace placeholders with findings as you query each tool. 3. Use ALL relevant tools - Query multiple databases for each data type; cross-reference across sources.

2. Citation Requirements (MANDATORY)

Every piece of information MUST include its source. Use inline citations:

*Source: PubChem via `PubChem_get_compound_properties_by_CID` (CID: 4091)*

3. Progressive Writing Workflow

Step 1:  Create report file with all section headers
Step 2:  Resolve compound identifiers -> Update Section 1
Step 3:  Query PubChem/ADMET-AI/DailyMed SPL -> Update Section 2 (Chemistry)
Step 4:  Query FDA Label MOA + ChEMBL + DGIdb -> Update Section 3 (Mechanism)
Step 5:  Query ADMET-AI tools -> Update Section 4 (ADMET)
Step 6:  Query ClinicalTrials.gov -> Update Section 5 (Clinical)
Step 7:  Query FAERS/DailyMed -> Update Section 6 (Safety)
Step 8:  Query PharmGKB -> Update Section 7 (Pharmacogenomics)
Step 9:  Query DailyMed/Orange Book -> Update Section 8 (Regulatory)
Step 10: Query PubMed/literature -> Update Section 9 (Literature)
Step 11: Synthesize findings -> Update Executive Summary & Section 10
Step 12: Document all sources -> Update Section 11 (Data Sources)

---

Compound Disambiguation (Phase 1)

CRITICAL: Establish compound identity before any research.

Identifier Resolution Chain

1. PubChem_get_CID_by_compound_name(compound_name)
   -> Extract: CID, canonical SMILES, formula

2. ChEMBL_search_molecules(query=drug_name)
   -> Extract: ChEMBL ID, pref_name

3. DailyMed_search_spls(drug_name)
   -> Extract: Set ID, NDC codes (if approved)

4. PharmGKB_search_drugs(query=drug_name)
   -> Extract: PharmGKB ID (PA...)

Handle Naming Ambiguity

---

Research Paths Summary

Each path has detailed tool chains and output examples in REPORT_GUIDELINES.md.

PATH 1: Chemical Properties & CMC

Tools: PubChem properties -> ADMET-AI physicochemical -> ADMET-AI solubility -> DailyMed chemistry/description Output: Physicochemical table, Lipinski assessment, QED score, salt forms, formulation comparison

PATH 2: Mechanism & Targets

Tools: DailyMed MOA -> ChEMBL activities (NOT ChEMBL_get_molecule_targets) -> ChEMBL target details -> DGIdb -> PubChem bioactivity Critical: Derive targets from activities filtered to pChEMBL >= 6.0. Avoid ChEMBL_get_molecule_targets. Output: FDA MOA text, target table with UniProt/potency, selectivity profile

PATH 3: ADMET Properties

Tools: ADMET-AI (bioavailability, BBB, CYP, clearance, toxicity) Fallback: DailyMed clinical_pharmacology + pharmacokinetics + drug_interactions Critical: If ADMET-AI fails, automatically use fallback. Never leave Section 4 empty.

PATH 4: Clinical Trials

Tools: search_clinical_trials -> compute phase counts -> extract outcomes/AEs -> fda_pharmacogenomic_biomarkers Critical: Section 5.2 must show actual counts by phase/status in table format.

PATH 5: Post-Marketing Safety

Tools: FAERS (reactions, seriousness, outcomes, deaths, age) + DailyMed (DDI, dosing, warnings) Critical: Include FAERS date window, seriousness breakdown, and limitations paragraph.

PATH 6: Pharmacogenomics

Tools: PharmGKB (search -> details -> annotations -> guidelines) Fallback: DailyMed pharmacogenomics section + PubMed literature

PATH 7: Regulatory & Patents

Tools: FDA Orange Book (search, approval history, exclusivity, patents, generics) + DailyMed (special populations via LOINC codes) Note: US-only data; document EMA/PMDA limitation.

PATH 8: Real-World Evidence

Tools: ClinicalTrials.gov (OBSERVATIONAL studies) + PubMed (real-world, registry, surveillance)

PATH 9: Comparative Analysis

Tools: Abbreviated tool chains for each comparator + head-to-head trial search + PubMed meta-analyses

---

FDA Label Core Fields

For approved drugs, retrieve these DailyMed sections early (after getting set_id):

---

Fallback Chains

---

Quick Reference: Tools by Use Case

See TOOLS_REFERENCE.md for the complete tool listing with parameters and input format requirements.

---

Type Normalization

Many tools require string inputs. Always convert IDs before API calls:

• ChEMBL IDs, PubMed IDs, NCT IDs: convert int -> str
• SMILES for ADMET-AI: pass as list ["SMILES_STRING"]
• FAERS drug names: use UPPERCASE (e.g., "METFORMIN")
• ChEMBL IDs: full format "CHEMBL1431" not "1431"
• PharmGKB IDs: PA prefix "PA450657" not "450657"

---

Common Use Cases

Always maintain all section headers but adjust depth based on query focus and data availability.

---

When NOT to Use This Skill

• Target research -> Use target-intelligence-gatherer skill
• Disease research -> Use disease-research skill
• Literature-only -> Use literature-deep-research skill
• Single property lookup -> Call tool directly
• Structure similarity search -> Use PubChem_search_compounds_by_similarity directly

---

Cross-Skill References

For drug interaction checking, run: python3 skills/tooluniverse-drug-drug-interaction/scripts/pharmacology_ref.py --type interaction --drug1 X --drug2 Y

---

Additional Resources

• Report template: REPORT_TEMPLATE.md - Initial file template, citation format, evidence grading, scorecard, audit template
• Report guidelines: REPORT_GUIDELINES.md - Detailed section-by-section instructions with output examples
• Tool reference: TOOLS_REFERENCE.md - Complete tool listing with parameters and input formats
• Verification checklist: CHECKLIST.md - Section-by-section pre-delivery verification
• Examples: EXAMPLES.md - Detailed workflow examples for different use cases

Drug Research Report Checklist

Complete verification checklist before delivering any drug research report.

---

Pre-Research Setup

Report File Created

• [ ] File named [DRUG]_drug_report.md
• [ ] All 11 section headers present
• [ ] [Researching...] placeholders in each section
• [ ] Header includes: Generated date, Query, Status

Compound Disambiguated

• [ ] PubChem CID established
• [ ] ChEMBL ID cross-referenced (or confirmed N/A)
• [ ] Canonical SMILES captured
• [ ] Naming ambiguity resolved (salt forms, isomers)
• [ ] All relevant identifiers documented

---

Section-by-Section Verification

Executive Summary

• [ ] 2-3 paragraphs synthesizing key findings
• [ ] Covers: What is it? What's it used for? Key characteristics?
• [ ] Bottom Line: One actionable sentence
• [ ] Written LAST (after all data collected)

Section 1: Compound Identity

Section 2: Chemical Properties

Section 3: Mechanism & Targets

Section 4: ADMET Properties

Section 5: Clinical Development

Section 6: Safety Profile

Section 7: Pharmacogenomics

Section 8: Regulatory & Labeling

Section 9: Literature & Research

Section 10: Conclusions & Assessment

Section 11: Data Sources & Methodology

---

Quality Checks

Evidence Grading Applied

• [ ] Clinical trial evidence marked ★★★
• [ ] Limited clinical evidence marked ★★☆
• [ ] Preclinical evidence marked ★☆☆
• [ ] Computational/indirect evidence marked ☆☆☆
• [ ] No ungraded claims in mechanism/efficacy sections

Citation Coverage

• [ ] Every table has source citation below
• [ ] Every factual claim has inline or block citation
• [ ] All tool names mentioned with parameters used
• [ ] Database IDs included (CID, ChEMBL ID, NCT, PMID)

Data Completeness Tier

• [ ] ●●● Complete: All 11 sections with substantial data
• [ ] ●●○ Substantial: 8-10 sections substantial, others noted sparse
• [ ] ●○○ Basic: Identity + chemistry + partial pharmacology
• [ ] ○○○ Minimal: Identity only (flag to user)

Formatting

• [ ] All tables render correctly
• [ ] No [Researching...] placeholders remaining
• [ ] Links work (PubChem, ChEMBL, ClinicalTrials.gov)
• [ ] Consistent formatting throughout
• [ ] Section numbers match template

---

Pre-Delivery Final Check

• [ ] Executive summary reflects complete report
• [ ] Bottom line is actionable and specific
• [ ] No search process text visible in report
• [ ] No raw tool outputs in report
• [ ] All 11 sections present
• [ ] Data sources section complete
• [ ] Report file saved
• [ ] Optional: [drug]_data.json created with structured data

---

Quick Fixes for Common Issues

Drug Research Examples

Detailed workflow examples demonstrating the drug research skill.

---

Example 1: Approved Drug Profile (Metformin)

User Query

"Tell me about metformin"

Step 1: Create Report File

Create metformin_drug_report.md with all section headers and [Researching...] placeholders.

Step 2: Compound Disambiguation

# Resolve identifiers
cid_result = tu.tools.PubChem_get_CID_by_compound_name(compound_name="metformin")
# → CID: 4091

chembl_result = tu.tools.ChEMBL_search_molecules(query="metformin", limit=1)
# → CHEMBL1431

dailymed_result = tu.tools.DailyMed_search_spls(drug_name="metformin")
# → Set ID: multiple entries (branded/generic)

pharmgkb_result = tu.tools.PharmGKB_search_drugs(query="metformin")
# → PA450657

Update Section 1 with:

## 1. Compound Identity

### 1.1 Database Identifiers
| Identifier | Value | Link |
|------------|-------|------|
| **PubChem CID** | 4091 | [Link](https://pubchem.ncbi.nlm.nih.gov/compound/4091) |
| **ChEMBL ID** | CHEMBL1431 | [Link](https://www.ebi.ac.uk/chembl/compound_report_card/CHEMBL1431/) |
| **RxNorm CUI** | 6809 | RxNorm |
| **PharmGKB ID** | PA450657 | [Link](https://www.pharmgkb.org/chemical/PA450657) |
| **DailyMed** | Multiple | FDA approved |

*Source: `PubChem_get_CID_by_compound_name`, `ChEMBL_search_drugs`, `PharmGKB_search_drugs`*

Step 3: Chemical Properties

# PubChem properties
props = tu.tools.PubChem_get_compound_properties_by_CID(cid=4091)
# → MW: 129.16, Formula: C4H11N5, XLogP: -2.6

# ADMET-AI predictions
smiles = "CN(C)C(=N)NC(=N)N"
physchem = tu.tools.ADMETAI_predict_physicochemical_properties(smiles=[smiles])
# → logP, TPSA, HBD, HBA, Lipinski, QED

solubility = tu.tools.ADMETAI_predict_solubility_lipophilicity_hydration(smiles=[smiles])
# → Solubility prediction

Update Section 2 with detailed tables and interpretations.

Step 4: Mechanism & Targets

# ChEMBL bioactivity
bioactivity = tu.tools.ChEMBL_search_activities(chembl_id="CHEMBL1431")
# → Activity data

targets = tu.tools.ChEMBL_get_target(chembl_id="CHEMBL1431")
# → Target list with UniProt

# DGIdb drug-gene interactions
dgidb = tu.tools.DGIdb_get_drug_info(drugs=["metformin"])
# → Gene interactions

Update Section 3 with target table, mechanism description, bioactivity summary.

Step 5: ADMET Properties

# Run all ADMET predictions
absorption = tu.tools.ADMETAI_predict_bioavailability(smiles=[smiles])
bbb = tu.tools.ADMETAI_predict_BBB_penetrance(smiles=[smiles])
cyp = tu.tools.ADMETAI_predict_CYP_interactions(smiles=[smiles])
clearance = tu.tools.ADMETAI_predict_clearance_distribution(smiles=[smiles])
toxicity = tu.tools.ADMETAI_predict_toxicity(smiles=[smiles])

Update Section 4 with ADMET tables:

### 4.1 Absorption
| Endpoint | Prediction | Interpretation |
|----------|------------|----------------|
| **Oral Bioavailability** | 0.68 | Moderate-Good |
| **HIA** | 0.92 | High absorption |
| **Caco-2** | -5.1 log cm/s | Moderate permeability |
| **P-gp Substrate** | 0.15 | Not a P-gp substrate |

*Source: ADMET-AI via `ADMETAI_predict_bioavailability`*

### 4.3 Metabolism
| CYP Enzyme | Inhibitor | Substrate |
|------------|-----------|-----------|
| CYP1A2 | 0.08 (No) | N/A |
| CYP2C9 | 0.05 (No) | 0.12 (No) |
| CYP2C19 | 0.07 (No) | N/A |
| CYP2D6 | 0.04 (No) | 0.08 (No) |
| CYP3A4 | 0.06 (No) | 0.15 (No) |

**Summary**: Metformin is not significantly metabolized by CYP enzymes. Eliminated unchanged renally.

*Source: ADMET-AI via `ADMETAI_predict_CYP_interactions`*

Step 6: Clinical Trials

# Search trials
trials = tu.tools.search_clinical_trials(intervention="metformin", pageSize=100)
# → Total: 500+ trials

# Get trial details for top results
nct_ids = [t['NCT ID'] for t in trials['studies'][:10]]
details = tu.tools.get_clinical_trial_conditions_and_interventions(nct_ids=nct_ids)

# Extract outcomes from completed Phase 3
outcomes = tu.tools.extract_clinical_trial_outcomes(
    nct_ids=["NCT00123456", "NCT00234567"],
    outcome_measure="primary"
)

Update Section 5 with trial landscape table, approved indications, key efficacy data.

Step 7: Safety Profile

# FAERS data
faers_reactions = tu.tools.FAERS_count_reactions_by_drug_event(
    medicinalproduct="METFORMIN"
)
# → Top adverse reactions

faers_serious = tu.tools.FAERS_count_seriousness_by_drug_event(
    medicinalproduct="METFORMIN"
)
# → Serious vs non-serious

faers_outcomes = tu.tools.FAERS_count_outcomes_by_drug_event(
    medicinalproduct="METFORMIN"
)
# → Outcome distribution

faers_deaths = tu.tools.FAERS_count_death_related_by_drug(
    medicinalproduct="METFORMIN"
)
# → Fatal outcomes

Update Section 6 with:

### 6.2 Post-Marketing Safety (FAERS)

**Total Reports**: 45,234 (as of 2026-02-04)

#### Top Adverse Reactions
| Reaction (MedDRA PT) | Count | % of Reports | Expected? |
|----------------------|-------|--------------|-----------|
| Diarrhoea | 8,234 | 18.2% | Yes (known GI effect) |
| Nausea | 6,892 | 15.2% | Yes (known GI effect) |
| Lactic acidosis | 3,456 | 7.6% | Yes (boxed warning) |
| Vomiting | 2,987 | 6.6% | Yes (known GI effect) |
| Hypoglycaemia | 2,543 | 5.6% | Yes (class effect) |

*Source: FDA FAERS via `FAERS_count_reactions_by_drug_event`*

#### Severity Distribution
| Classification | Count | Percentage |
|----------------|-------|------------|
| Serious | 18,456 | 40.8% |
| Non-serious | 26,778 | 59.2% |

*Source: `FAERS_count_seriousness_by_drug_event`*

### 6.3 Black Box Warnings

**Lactic Acidosis Warning**: Metformin can cause lactic acidosis, a rare but serious 
metabolic complication. Risk increases with renal impairment, hepatic impairment, 
age ≥65, radiologic studies with contrast, surgery, excessive alcohol intake, and 
hypoxic states.

*Source: FDA Label via `FDA_get_drug_label`*

Step 8: Pharmacogenomics

# PharmGKB data
pgkb_drug = tu.tools.PharmGKB_search_drugs(query="metformin")
pgkb_details = tu.tools.PharmGKB_get_drug_details(drug_id="PA450657")

# Clinical annotations for related genes
annotations = tu.tools.PharmGKB_get_clinical_annotations(gene_id="PA35858")  # SLC22A1

# Dosing guidelines
guidelines = tu.tools.PharmGKB_get_dosing_guidelines(gene="SLC22A1")

Update Section 7 with pharmacogene table, clinical annotations, guideline status.

Step 9: Literature

# PubMed search
lit_total = tu.tools.PubMed_search_articles(
    query='"metformin"',
    limit=1
)
# → Total count

lit_recent = tu.tools.PubMed_search_articles(
    query='"metformin" AND ("2024"[Date - Publication] OR "2025"[Date - Publication])',
    limit=20
)
# → Recent papers

lit_drug = tu.tools.PubMed_search_articles(
    query='"metformin" AND (drug OR therapy OR treatment)',
    limit=20
)
# → Drug-focused papers

Update Section 9 with publication metrics and research trends.

Step 10: Synthesize Conclusions

Update Section 10 with scorecard:

### 10.1 Drug Profile Scorecard

| Criterion | Score | Rationale |
|-----------|-------|-----------|
| **Efficacy Evidence** | 5/5 | Decades of clinical use, multiple Phase 3 trials, proven HbA1c reduction |
| **Safety Profile** | 4/5 | Generally well-tolerated; lactic acidosis is rare but serious (boxed warning) |
| **PK/ADMET** | 4/5 | Good oral bioavailability; renal elimination (caution in CKD) |
| **Target Validation** | 4/5 | AMPK activation well-established; some mechanism debate remains |
| **Competitive Position** | 3/5 | First-line for T2DM but many alternatives (SGLT2i, GLP-1) |
| **Overall** | **4.0/5** | **Strong, established drug profile** |

### 10.2 Key Strengths
1. **Proven efficacy**: 60+ years of clinical use with robust trial data
2. **Favorable safety**: Low hypoglycemia risk, potential cardiovascular benefits
3. **Cost-effective**: Generic availability, low cost
4. **Additional benefits**: Weight neutral/loss, potential anti-cancer properties under investigation

### 10.3 Key Concerns/Limitations
1. **Lactic acidosis risk**: Requires monitoring, contraindicated in renal impairment
2. **GI tolerability**: ~25% of patients experience GI side effects
3. **B12 deficiency**: Long-term use associated with vitamin B12 malabsorption
4. **Renal clearance**: Dose adjustment required for eGFR <45

### 10.4 Research Gaps
1. **Cancer prevention**: Clinical trials ongoing but definitive evidence lacking
2. **Aging/longevity**: TAME trial results pending

Step 11: Update Data Sources

Update Section 11 with all tools used and their parameters.

Final Step: Write Executive Summary

## Executive Summary

**Metformin** (PubChem CID: 4091, ChEMBL: CHEMBL1431) is a biguanide antidiabetic 
agent and the first-line pharmacological treatment for type 2 diabetes mellitus. 
It works primarily through AMPK activation, reducing hepatic glucose production 
and improving insulin sensitivity.

The drug has an extensive clinical evidence base spanning 60+ years with 500+ 
registered clinical trials. Key efficacy data shows HbA1c reductions of 1.0-1.5% 
as monotherapy [★★★]. The safety profile is generally favorable, though a boxed 
warning exists for lactic acidosis risk in patients with renal impairment. FAERS 
data (45,234 reports) shows GI effects as the most common adverse reactions, 
consistent with labeling.

Pharmacogenomic data from PharmGKB identifies SLC22A1 (OCT1) as the key 
transporter affecting metformin response, though no CPIC dosing guideline 
currently exists. The drug shows excellent ADMET properties with good 
oral bioavailability and minimal CYP-mediated metabolism.

**Bottom Line**: Metformin remains a well-validated, cost-effective first-line 
therapy for T2DM with a favorable benefit-risk profile when renal function is 
monitored. Active research continues into potential benefits in cancer prevention 
and longevity.

---

Example 2: Investigational Compound (ChEMBL ID)

User Query

"What do we know about compound CHEMBL4303291?"

Approach Differences

1. More emphasis on preclinical data - Chemistry, targets, ADMET predictions 2. Clinical section may be sparse - No or few trials 3. No FAERS data - Not yet marketed 4. No PGx data - Not yet characterized 5. Label section: "Not approved - investigational compound"

Key Tool Adjustments

# Start with ChEMBL ID directly
compound = tu.tools.ChEMBL_get_compound_by_chemblid(chembl_id="CHEMBL4303291")
smiles = compound['molecule_structures']['canonical_smiles']

# Get PubChem CID from SMILES
cid_result = tu.tools.PubChem_get_CID_by_SMILES(smiles=smiles)

# Heavy emphasis on ADMET predictions (no clinical data)
# All 8 ADMET-AI tools...

# Search for any trials
trials = tu.tools.search_clinical_trials(intervention="CHEMBL4303291")
# → Likely 0 results

# Literature may be primary data source
lit = tu.tools.PubMed_search_articles(query='"CHEMBL4303291"', limit=20)

Report Differences

• Section 5 (Clinical): "No clinical trials registered as of [date]. Compound is in preclinical development."
• Section 6 (Safety): "No post-marketing safety data (not approved). Preclinical toxicity data: [from ChEMBL/literature]"
• Section 7 (PGx): "No pharmacogenomic data available (investigational compound)"
• Section 8 (Regulatory): "Not approved. Development stage: Preclinical"
• Scorecard: Adjust criteria for investigational compound assessment

---

Example 3: Safety-Focused Review

User Query

"What are the safety concerns with fluoroquinolone antibiotics like ciprofloxacin?"

Approach Differences

1. Deep FAERS analysis - Multiple queries, stratified analysis 2. Black box warnings emphasis - Known serious effects 3. Drug interactions detailed - QT prolongation, etc. 4. PGx for safety - Variants affecting toxicity 5. Literature focused on safety - Adverse event publications

Key Tool Sequence

# Full FAERS analysis
reactions = tu.tools.FAERS_count_reactions_by_drug_event(
    medicinalproduct="CIPROFLOXACIN"
)

# Filter for serious reactions
serious_cardiac = tu.tools.FAERS_count_reactions_by_drug_event(
    medicinalproduct="CIPROFLOXACIN",
    serious="Yes",
    reactionmeddraverse="cardiac"  # Will need specific MedDRA terms
)

# Age stratification
age_dist = tu.tools.FAERS_count_patient_age_distribution(
    medicinalproduct="CIPROFLOXACIN"
)

# Death-related
deaths = tu.tools.FAERS_count_death_related_by_drug(
    medicinalproduct="CIPROFLOXACIN"
)

# Label for black box warnings
# FDA labels are keyed by drug name, not CID -- resolve the name first
_syn = tu.tools.PubChem_get_compound_synonyms_by_CID(cid=2764)
_name = _syn['data'][0] if isinstance(_syn, dict) and _syn.get('data') else None
label = tu.tools.FDA_get_drug_label(drug_name=_name)

# Literature on safety
safety_lit = tu.tools.PubMed_search_articles(
    query='"ciprofloxacin" AND (safety OR "adverse event" OR toxicity OR "side effect")',
    limit=50
)

Report Emphasis

• Section 6 expanded with multiple subsections for:
• Tendon rupture/tendinitis
• QT prolongation
• CNS effects
• Peripheral neuropathy
• Aortic dissection/aneurysm
• Safety scorecard instead of general scorecard
• Risk mitigation strategies section added

---

Example 4: ADMET Assessment (SMILES Input)

User Query

"Evaluate this compound's drug-likeness: CC(=O)Nc1ccc(O)cc1"

Approach

1. Identify the compound first 2. Heavy chemistry/ADMET focus 3. Minimal clinical sections (unless identified as known drug)

Workflow

smiles = "CC(=O)Nc1ccc(O)cc1"  # Paracetamol/Acetaminophen

# Try to identify
cid = tu.tools.PubChem_get_CID_by_SMILES(smiles=smiles)
# → CID: 1983 (Acetaminophen)

# If identified, proceed with full profile
# If not identified, focus on predictions only

# Full ADMET battery
physchem = tu.tools.ADMETAI_predict_physicochemical_properties(smiles=[smiles])
bioavail = tu.tools.ADMETAI_predict_bioavailability(smiles=[smiles])
bbb = tu.tools.ADMETAI_predict_BBB_penetrance(smiles=[smiles])
cyp = tu.tools.ADMETAI_predict_CYP_interactions(smiles=[smiles])
clearance = tu.tools.ADMETAI_predict_clearance_distribution(smiles=[smiles])
toxicity = tu.tools.ADMETAI_predict_toxicity(smiles=[smiles])
nuclear = tu.tools.ADMETAI_predict_nuclear_receptor_activity(smiles=[smiles])
stress = tu.tools.ADMETAI_predict_stress_response(smiles=[smiles])

Report Focus

• Sections 2 and 4 are primary - Detailed ADMET analysis
• Drug-likeness verdict prominent - Lead-like, drug-like, or beyond Rule of 5
• Structure-activity insights if related compounds found
• Other sections: Brief if compound identified; "N/A - SMILES query" if unknown

---

Example 5: Clinical Development Landscape

User Query

"What trials are ongoing for KRAS inhibitors?"

Approach

1. ClinicalTrials.gov focused 2. Multiple drugs in class - Not single compound 3. Competitive landscape emphasis 4. Pipeline analysis

Workflow

# Search for KRAS inhibitors
kras_trials = tu.tools.search_clinical_trials(
    condition="cancer",
    intervention="KRAS",
    pageSize=100
)

# Specific drugs
sotorasib_trials = tu.tools.search_clinical_trials(intervention="sotorasib")
adagrasib_trials = tu.tools.search_clinical_trials(intervention="adagrasib")

# Get trial details
nct_ids = [t['NCT ID'] for t in kras_trials['studies'][:20]]
conditions = tu.tools.get_clinical_trial_conditions_and_interventions(nct_ids=nct_ids)
status = tu.tools.get_clinical_trial_status_and_dates(nct_ids=nct_ids)

# Outcomes from completed trials
completed = [nct for nct in nct_ids if "Completed" in status[nct]]
outcomes = tu.tools.extract_clinical_trial_outcomes(nct_ids=completed)

Report Differences

• Class-level analysis rather than single drug
• Competitive landscape table - All KRAS inhibitors in development
• Indication mapping - Which cancers, which mutations
• Pipeline by phase chart
• Less emphasis on individual ADMET/chemistry (covered per drug if needed)

---

Summary: Adjusting by Use Case

Always maintain all section headers but adjust depth based on query focus and data availability.

Drug Research Report Guidelines

Detailed section-by-section instructions with output examples for each research path.

---

Report Detail Requirements

Each section must be comprehensive and detailed:

• Tables: Use tables for structured data (targets, trials, adverse events)
• Lists: Use bullet points for features, findings, key points
• Paragraphs: Include narrative summaries that synthesize findings
• Numbers: Include specific values, counts, percentages (not vague terms)
• Context: Explain what the data means, not just what it is

BAD (too brief):

### Clinical Trials
Multiple trials completed. Approved for diabetes.

GOOD (detailed with sources):

### 5.2 Clinical Trial Landscape
| Phase | Total | Completed | Recruiting |
|-------|-------|-----------|------------|
| Phase 4 | 89 | 72 | 12 |
| Phase 3 | 156 | 134 | 15 |

**Total Registered Trials**: 515 (as of 2026-02-04)
*Source: ClinicalTrials.gov via `search_clinical_trials`*

---

FDA Label Core Fields Bundle

For approved drugs, ALWAYS retrieve these FDA label sections early (after getting set_id from DailyMed_search_spls).

Call DailyMed_get_spl_by_setid(setid=set_id, sections=[...]) in batches:

Phase 1 (Mechanism & Chemistry):

• mechanism_of_action -> Section 3.1
• pharmacodynamics -> Section 3.1
• chemistry -> Section 2.4

Phase 2 (ADMET & PK):

• clinical_pharmacology -> Section 4
• pharmacokinetics -> Section 4.1-4.4
• drug_interactions -> Section 4.3, 6.5

Phase 3 (Safety & Dosing):

• warnings_and_cautions -> Section 6.3
• adverse_reactions -> Section 6.1
• dosage_and_administration -> Section 6.6, 8.2

Phase 4 (PGx & Clinical):

• pharmacogenomics -> Section 7
• clinical_studies -> Section 5.5
• description -> Section 2.5 (formulation)
• inactive_ingredients -> Section 2.5

---

PATH 1: Chemical Properties & CMC

Objective: Full physicochemical profile, salt forms, formulation details.

Tool Chain: 1. PubChem_get_compound_properties_by_CID(cid) -> MW, formula, XLogP, TPSA, HBD, HBA, rotatable bonds 2. ADMETAI_predict_physicochemical_properties(smiles=[smiles]) -> logP, Lipinski, QED, TPSA 3. ADMETAI_predict_solubility_lipophilicity_hydration(smiles=[smiles]) -> Solubility, Lipophilicity 4. DailyMed_get_spl_by_setid(setid, sections=["chemistry"]) -> Salt forms, polymorphs 5. DailyMed_get_spl_by_setid(setid, sections=["description", "inactive_ingredients"]) -> Formulation

Type Normalization: Convert all numeric IDs to strings before API calls.

Output Example (Section 2.1):

| Property | Value | Drug-Likeness | Source |
|----------|-------|---------------|--------|
| **Molecular Weight** | 129.16 g/mol | Pass (< 500) | PubChem |
| **LogP** | -2.64 | Pass (< 5) | ADMET-AI |
| **TPSA** | 91.5 A^2 | Pass (< 140) | PubChem |
| **H-Bond Donors** | 2 | Pass (<= 5) | PubChem |

**Lipinski Rule of Five**: PASS (0 violations)
**QED Score**: 0.74 (Good drug-likeness)
*Sources: PubChem, ADMET-AI*

Formulation Comparison (if multiple formulations exist):

| Formulation | Tmax (h) | Cmax (ng/mL) | AUC | Half-life (h) | Dosing |
|-------------|----------|--------------|-----|---------------|--------|
| **IR** | 2.5 | 1200 | 8400 | 6.5 | 500 mg TID |
| **ER** | 7.0 | 950 | 8900 | 6.5 | 1000 mg QD |
*Source: DailyMed clinical pharmacology sections*

---

PATH 2: Mechanism & Targets

Objective: FDA label MOA + experimental targets + selectivity.

Tool Chain: 1. DailyMed_get_spl_by_setid(setid, sections=["mechanism_of_action", "pharmacodynamics"]) -> Official FDA MOA [T1] 2. ChEMBL_search_activities(molecule_chembl_id=chembl_id, limit=100) -> Activity records 3. ChEMBL_get_target(target_chembl_id) for each unique target -> Target name, UniProt [T1] 4. DGIdb_get_drug_info(drugs=[drug_name]) -> Target genes, interaction types [T2] 5. PubChem_get_compound_bioactivity(cid) -> Assay summary [T2]

CRITICAL:

• Avoid `ChEMBL_get_molecule_targets` - returns unfiltered/irrelevant targets
• Derive targets from activities: Filter to pChEMBL >= 6.0 or IC50/EC50 <= 1 uM
• Type normalization: Convert all ChEMBL IDs to strings

Output Example (Section 3.2):

| Target | UniProt | Type | Potency | Assays | Evidence | Source |
|--------|---------|------|---------|--------|----------|--------|
| PRKAA1 (AMPK a1) | Q13131 | Activator | EC50 ~10 uM | 12 | T1 | ChEMBL |
| SLC22A1 (OCT1) | O15245 | Substrate | Km ~1.5 mM | 5 | T2 | DGIdb |
*Source: ChEMBL activities filtered to pChEMBL >= 6.0*

---

PATH 3: ADMET Properties

Objective: Full ADMET profile - predictions + FDA label PK.

Primary Chain (ADMET-AI): 1. ADMETAI_predict_bioavailability(smiles) -> Bioavailability, HIA, PAMPA, Caco2, Pgp 2. ADMETAI_predict_BBB_penetrance(smiles) -> BBB probability 3. ADMETAI_predict_CYP_interactions(smiles) -> CYP1A2, 2C9, 2C19, 2D6, 3A4 4. ADMETAI_predict_clearance_distribution(smiles) -> Clearance, Half_Life, VDss, PPBR 5. ADMETAI_predict_toxicity(smiles) -> AMES, hERG, DILI, ClinTox, LD50

Fallback Chain (if ADMET-AI fails): 1. DailyMed_get_spl_by_setid(setid, sections=["clinical_pharmacology", "pharmacokinetics"]) [T1] 2. DailyMed_get_spl_by_setid(setid, sections=["drug_interactions"]) [T1] 3. PubMed_search_articles(query="[drug] pharmacokinetics", max_results=10) [T2]

CRITICAL: If ADMET-AI tools fail, automatically switch to fallback. Do NOT leave Section 4 as "predictions unavailable."

---

PATH 4: Clinical Trials

Objective: Complete clinical development picture with accurate phase counts.

Tool Chain: 1. search_clinical_trials(intervention=drug_name, pageSize=100) -> Full result set 2. COMPUTE PHASE COUNTS from results (Phase 1/2/3/4 by status) 3. SELECT REPRESENTATIVE TRIALS (top 5 Phase 3, top 3 recruiting) 4. get_clinical_trial_conditions_and_interventions(nct_ids=selected) -> Details 5. extract_clinical_trial_outcomes(nct_ids=completed_phase3) -> Efficacy 6. extract_clinical_trial_adverse_events(nct_ids=completed) -> Safety 7. fda_pharmacogenomic_biomarkers(drug_name) -> Companion diagnostics [T1] 8. PharmGKB_get_clinical_annotations(drug_id) -> Response predictors [T2]

CRITICAL: Section 5.2 must show actual counts by phase/status in table format, not just a trial list.

Output Example (Section 5.6 - Biomarkers):

#### FDA-Required Testing
| Biomarker | Requirement Level | Approved Test(s) | Evidence |
|-----------|------------------|------------------|----------|
| EGFR T790M | Required (NSCLC) | cobas v2 | T1 |
*Source: `fda_pharmacogenomic_biomarkers`*

---

PATH 5: Post-Marketing Safety & Drug Interactions

Objective: Real-world safety signals + DDI guidance + dose modifications.

FAERS Chain: 1. FAERS_count_reactions_by_drug_event(medicinalproduct) -> Top 20 AEs [T1] 2. FAERS_count_seriousness_by_drug_event(medicinalproduct) -> Serious ratio [T1] 3. FAERS_count_outcomes_by_drug_event(medicinalproduct) -> Outcomes [T1] 4. FAERS_count_death_related_by_drug(medicinalproduct) -> Fatal count [T1] 5. FAERS_count_patient_age_distribution(medicinalproduct) -> Age groups [T1]

DDI & Dose Modification Chain: 6. DailyMed_get_spl_by_setid(setid, sections=["drug_interactions"]) -> DDI table [T1] 7. DailyMed_get_spl_by_setid(setid, sections=["dosage_and_administration", "warnings_and_cautions"]) -> Dose mods [T1] 8. DailyMed_get_spl_by_setid(setid) -> Drug-food interactions (search for grapefruit, alcohol, food, meal) [T1] 9. search_clinical_trials(intervention="[drug] AND combination") -> Approved combos [T1]

CRITICAL FAERS Requirements:

• Include date window (e.g., "Reports from 2004-2026")
• Report seriousness breakdown (not just top PTs)
• Add limitations paragraph: Small N, voluntary reporting, causality not established, reporting bias

Output Example (Section 6.6 - Dose Modifications):

#### Renal Impairment
| eGFR (mL/min/1.73m^2) | Dosing |
|----------------------|--------|
| >=60 | No adjustment |
| 45-59 | Max 1000 mg/day |
| 30-44 | Max 500 mg/day |
| <30 | Contraindicated |
*Source: DailyMed SPL*

---

PATH 6: Pharmacogenomics

Primary Chain (PharmGKB): 1. PharmGKB_search_drugs(query) -> PharmGKB drug ID 2. PharmGKB_get_drug_details(drug_id) -> Cross-references, related genes 3. PharmGKB_get_clinical_annotations(gene_id) for each gene -> Variant-drug associations 4. PharmGKB_get_dosing_guidelines(gene=gene_symbol) -> CPIC/DPWG recommendations

Fallback Chain: 5. DailyMed_get_spl_by_setid(setid, sections=["pharmacogenomics", "clinical_pharmacology"]) [T1] 6. PubMed_search_articles(query="[drug] pharmacogenomics", max_results=5) [T2]

---

PATH 7: Regulatory Status & Patents

Tool Chain: 1. DailyMed_search_spls(drug_name) -> SetID 2. FDA_OrangeBook_search_drug(brand_name) -> Application number [T1] 3. FDA_OrangeBook_get_approval_history(appl_no) -> Approval dates [T1] 4. FDA_OrangeBook_get_exclusivity(brand_name) -> Exclusivity types/dates [T1] 5. FDA_OrangeBook_get_patent_info(brand_name) -> Patent numbers [T1] 6. FDA_OrangeBook_check_generic_availability(brand_name) -> Generic entries [T1] 7. DailyMed_get_spl_by_setid(setid) -> Special populations (pediatric, geriatric, pregnancy, lactation, renal, hepatic) using LOINC codes [T1]

CRITICAL: Orange Book data is US-only. Document limitation for EMA/PMDA.

---

PATH 8: Real-World Evidence

Tool Chain: 1. search_clinical_trials(study_type="OBSERVATIONAL", intervention=drug_name) -> RWE studies [T1] 2. PubMed_search_articles(query="[drug] (real-world OR observational OR effectiveness)") [T2] 3. PubMed_search_articles(query="[drug] (registry OR post-marketing OR surveillance)") [T2] 4. Compare clinical trial efficacy vs real-world effectiveness

---

PATH 9: Comparative Analysis

Tool Chain: 1. Identify comparator drugs (user-provided or inferred from class) 2. For each comparator: abbreviated tool chain (PubChem, ChEMBL potency, trials, FAERS) 3. search_clinical_trials(intervention="[drug] AND [comparator]") -> Head-to-head trials [T1] 4. PubMed_search_articles(query="[drug] vs [comparator]") -> Meta-analyses [T2] 5. Create comparison tables: potency, selectivity, ADMET, efficacy, safety

---

Type Normalization & Error Prevention

Many ToolUniverse tools require string inputs but may return integers/floats. Always convert IDs:

chembl_ids = [str(id) for id in chembl_ids]
nct_ids = [str(id) for id in nct_ids]
pmids = [str(id) for id in pmids]

Pre-call checklist:

• All ID parameters are strings
• Lists contain strings, not ints/floats
• No None/null values in required fields
• Arrays are non-empty if required

Drug Research Report Template

Initial report template to create before any tool calls. File name: [DRUG]_drug_report.md

---

Template

# Drug Research Report: [DRUG NAME]

**Generated**: [Date] | **Query**: [Original query] | **Status**: In Progress

---

## Executive Summary
[Researching...]

---

## 1. Compound Identity
### 1.1 Database Identifiers
[Researching...]
### 1.2 Structural Information
[Researching...]
### 1.3 Names & Synonyms
[Researching...]

---

## 2. Chemical Properties
### 2.1 Physicochemical Profile
[Researching...]
### 2.2 Drug-Likeness Assessment
[Researching...]
### 2.3 Solubility & Permeability
[Researching...]
### 2.4 Salt Forms & Polymorphs
[Researching...]
### 2.5 Structure Visualization
[Researching...]

---

## 3. Mechanism & Targets
### 3.1 Primary Mechanism of Action
[Researching...]
### 3.2 Primary Target(s)
[Researching...]
### 3.3 Target Selectivity & Off-Targets
[Researching...]
### 3.4 Bioactivity Profile (ChEMBL)
[Researching...]

---

## 4. ADMET Properties
### 4.1 Absorption
[Researching...]
### 4.2 Distribution
[Researching...]
### 4.3 Metabolism
[Researching...]
### 4.4 Excretion
[Researching...]
### 4.5 Toxicity Predictions
[Researching...]

---

## 5. Clinical Development
### 5.1 Development Status
[Researching...]
### 5.2 Clinical Trial Landscape
[Researching...]
### 5.3 Approved Indications
[Researching...]
### 5.4 Investigational Indications
[Researching...]
### 5.5 Key Efficacy Data
[Researching...]
### 5.6 Biomarkers & Companion Diagnostics
[Researching...]

---

## 6. Safety Profile
### 6.1 Clinical Adverse Events
[Researching...]
### 6.2 Post-Marketing Safety (FAERS)
[Researching...]
### 6.3 Black Box Warnings
[Researching...]
### 6.4 Contraindications
[Researching...]
### 6.5 Drug-Drug Interactions
[Researching...]
### 6.5.2 Drug-Food Interactions
[Researching...]
### 6.6 Dose Modification Guidance
[Researching...]
### 6.7 Drug Combinations & Regimens
[Researching...]

---

## 7. Pharmacogenomics
### 7.1 Relevant Pharmacogenes
[Researching...]
### 7.2 Clinical Annotations
[Researching...]
### 7.3 Dosing Guidelines (CPIC/DPWG)
[Researching...]
### 7.4 Actionable Variants
[Researching...]

---

## 8. Regulatory & Labeling
### 8.1 Approval Status
[Researching...]
### 8.2 Label Highlights
[Researching...]
### 8.3 Patents & Exclusivity
[Researching...]
### 8.4 Label Changes & Warnings
[Researching...]
### 8.5 Special Populations
[Researching...]
### 8.6 Regulatory Timeline & History
[Researching...]

---

## 9. Literature & Research Landscape
### 9.1 Publication Metrics
[Researching...]
### 9.2 Research Themes
[Researching...]
### 9.3 Recent Key Publications
[Researching...]
### 9.4 Real-World Evidence
[Researching...]

---

## 10. Conclusions & Assessment
### 10.1 Drug Profile Scorecard
[Researching...]
### 10.2 Key Strengths
[Researching...]
### 10.3 Key Concerns/Limitations
[Researching...]
### 10.4 Research Gaps
[Researching...]
### 10.5 Comparative Analysis
[Researching...]

---

## 11. Data Sources & Methodology
### 11.1 Primary Data Sources
[Researching...]
### 11.2 Tool Call Summary
[Researching...]
### 11.3 Quality Control Metrics
[Researching...]

Then progressively replace [Researching...] with actual findings as you query each tool.

---

Citation Format

For each data section, include source attribution:

*Source: PubChem via `PubChem_get_compound_properties_by_CID` (CID: 4091)*

Block citation at section end:

---
**Data Sources for this section:**
- PubChem: `PubChem_get_compound_properties_by_CID` (CID: 4091)
- ChEMBL: `ChEMBL_get_activity` (CHEMBL1431)
- DGIdb: `DGIdb_get_drug_info` (metformin)
---

---

Evidence Grading System

Apply inline: Metformin reduces hepatic glucose output via AMPK activation [T1: FDA Label].

Include per-section quality summary:

**Evidence Quality**: Strong (156 Phase 3 trials)
**Data Confidence**: High - mature clinical program

---

Drug Profile Scorecard (Section 10.1)

| Criterion | Score (1-5) | Rationale |
|-----------|-------------|-----------|
| **Efficacy Evidence** | 5 | Multiple Phase 3 trials, decades of use |
| **Safety Profile** | 4 | Well-tolerated; rare but serious risks |
| **PK/ADMET** | 4 | Good bioavailability; known clearance |
| **Target Validation** | 4 | Mechanism well-established |
| **Competitive Position** | 3 | First-line but alternatives exist |
| **Overall** | 4.0 | **Strong drug profile** |

**Interpretation**: 5 = Excellent, 4 = Good, 3 = Moderate, 2 = Concerning, 1 = Poor

---

Completeness Audit Template (Section 11)

## Report Completeness Audit

**Overall Completeness**: XX% (N/M minimum requirements met)

### Missing Data Items
| Section | Missing Item | Recommended Action |
|---------|--------------|-------------------|
| 2 | Salt forms | Call DailyMed chemistry section |

### Tool Failures Encountered
| Tool | Error | Fallback Used |
|------|-------|---------------|
| `PharmGKB_search_drugs` | API timeout | DailyMed label PGx sections |

### Data Confidence Assessment
| Section | Confidence | Evidence Tier |
|---------|-----------|---------------|
| 1. Identity | High | T1 |
| 4. ADMET | Medium | T2 |

### Cross-Source Validation
| Property | Source 1 | Source 2 | Agreement |
|----------|---------|---------|-----------|
| MW | 378.88 (PubChem) | 378.88 (ChEMBL) | Exact match |

### Evidence Distribution
| Tier | Count | Percentage |
|------|-------|------------|
| T1 | 45 | 65% |
| T2 | 18 | 26% |
| T3 | 5 | 7% |
| T4 | 1 | 1% |

Drug Research Tools Reference

Complete reference for 50+ tools used in drug research, organized by use case.

---

Quick Reference: Tools by Use Case

---

Compound Identity & Chemistry

PubChem Tools

ChEMBL Tools

---

ADMET Predictions

All ADMET-AI tools require SMILES input in list format: smiles=["CC(=O)Oc1ccccc1C(=O)O"]

Physicochemical

Absorption

Distribution

Metabolism

Excretion

Toxicity

---

Drug-Gene Interactions

DGIdb Tools

---

Clinical Trials (ClinicalTrials.gov)

Search & Overview

Details

Results Extraction

---

Adverse Events (FDA FAERS)

Single Drug Analysis

Multi-Drug Analysis

Filter Options (FAERS)

---

🆕 FAERS Analytics (Statistical Signal Detection)

NEW: Priority 1 tools for pharmacovigilance with regulatory-grade statistics.

Disproportionality Analysis

Example:

result = tu.tools.FAERS_calculate_disproportionality(
    operation="calculate_disproportionality",
    drug_name="IBUPROFEN",
    adverse_event="Gastrointestinal haemorrhage"
)
# Returns: ROR 2.8 [2.71-2.89], "Moderate signal"

Interpretation:

• ROR > 1: Positive association (higher reporting than expected)
• Signal detected: ROR lower CI > 1.0 AND case count ≥ 3
• Signal strength: Strong (ROR ≥4), Moderate (ROR ≥2), Weak (ROR <2)
• ⚠️ Caution: Disproportionality ≠ causation (association only)

Demographic Stratification

Example:

result = tu.tools.FAERS_stratify_by_demographics(
    operation="stratify_by_demographics",
    drug_name="IBUPROFEN",
    adverse_event="Gastrointestinal haemorrhage",
    stratify_by="sex"  # Options: sex, age, country
)
# Returns: Female 60%, Male 40%

Serious Events Filtering

Example:

result = tu.tools.FAERS_filter_serious_events(
    operation="filter_serious_events",
    drug_name="IBUPROFEN",
    seriousness_type="death"  # Options: all, death, hospitalization, disability, life_threatening
)
# Returns: Total deaths, top fatal reactions

Drug Comparison

Example:

result = tu.tools.FAERS_compare_drugs(
    operation="compare_drugs",
    drug1="IBUPROFEN",
    drug2="NAPROXEN",
    adverse_event="Gastrointestinal haemorrhage"
)
# Returns: Comparative analysis with ROR for both drugs

Temporal Trend Analysis

Example:

result = tu.tools.FAERS_analyze_temporal_trends(
    operation="analyze_temporal_trends",
    drug_name="IBUPROFEN",
    adverse_event="Gastrointestinal haemorrhage"
)
# Returns: Increasing/Decreasing/Stable trend with percent change

MedDRA Hierarchy

Example:

result = tu.tools.FAERS_rollup_meddra_hierarchy(
    operation="rollup_meddra_hierarchy",
    drug_name="IBUPROFEN"
)
# Returns: List of PTs with case counts

---

🆕 DailyMed SPL Parser (Structured Label Extraction)

NEW: Priority 1 tools for extracting structured data from FDA product labels.

Parse Operations

Requirements: Install lxml with pip install lxml

Example:

# First, get setid from search
search = tu.tools.DailyMed_search_spls(drug_name="ibuprofen")
setid = search['data']['results'][0]['setid']

# Parse adverse reactions
result = tu.tools.DailyMed_parse_adverse_reactions(
    operation="parse_adverse_reactions",
    setid=setid
)
# Returns: {"tables": [...], "text_sections": [...]}

Use Cases:

• Systematic AE frequency extraction (no more manual PDF parsing)
• Structured dose modification rules
• Contraindication screening
• Drug-drug interaction tables for clinical decision support

---

🆕 FDA Orange Book (US Regulatory Intelligence)

NEW: Priority 1 tools for FDA approval data, patents, and generic availability.

Search & Basic Info

Example:

result = tu.tools.FDA_OrangeBook_search_drug(
    operation="search_drug",
    brand_name="ADVIL"
)
# Or search by application number
result = tu.tools.FDA_OrangeBook_search_drug(
    operation="search_drug",
    application_number="NDA020402"
)

Regulatory Timeline

Example:

result = tu.tools.FDA_OrangeBook_get_approval_history(
    operation="get_approval_history",
    application_number="NDA020402"
)
# Returns: Approval date, submissions, review documents

Patent & Exclusivity

Example:

# Patent info (guidance + download URL)
patents = tu.tools.FDA_OrangeBook_get_patent_info(
    operation="get_patent_info",
    brand_name="ADVIL"
)

# Exclusivity info
exclusivity = tu.tools.FDA_OrangeBook_get_exclusivity(
    operation="get_exclusivity",
    brand_name="ADVIL"
)

Generic Availability

Example:

# Check generic availability
generics = tu.tools.FDA_OrangeBook_check_generic_availability(
    operation="check_generic_availability",
    brand_name="ADVIL"
)
# Returns: Reference product, generic count, TE codes

# Get TE code interpretation
te_codes = tu.tools.FDA_OrangeBook_get_te_code(
    operation="get_te_code",
    brand_name="ADVIL"
)
# Returns: TE codes (e.g., "AB") with interpretation

TE Code Guide:

• A: Therapeutically equivalent
• AB: Bioequivalence standards met
• AP: Multi-source with equivalence
• B: NOT therapeutically equivalent
• BC: Extended-release dosage form issues
• BN: Active ingredient issues

Use Cases:

• Market entry planning (patent cliffs, exclusivity expiry)
• Generic competition landscape
• Regulatory submission timelines
• Therapeutic equivalence assessment

---

Drug Labeling (Legacy)

DailyMed Tools

---

Pharmacogenomics (PharmGKB)

---

Literature

Search Tools

Citation Tools

---

Tool Chains by Research Path

PATH 1: Identity Resolution

PubChem_get_CID_by_compound_name
  ↓
ChEMBL_search_molecules
  ↓
DailyMed_search_spls
  ↓
PharmGKB_search_drugs

PATH 2: Chemistry & Drug-likeness

PubChem_get_compound_properties_by_CID
  ↓
ADMETAI_predict_physicochemical_properties
  ↓
ADMETAI_predict_solubility_lipophilicity_hydration

PATH 3: Targets & Bioactivity

ChEMBL_search_activities
  ↓
ChEMBL_get_target
  ↓
DGIdb_get_drug_info
  ↓
PubChem_get_compound_bioactivity

PATH 4: ADMET Profile

ADMETAI_predict_bioavailability
  ↓
ADMETAI_predict_BBB_penetrance
  ↓
ADMETAI_predict_CYP_interactions
  ↓
ADMETAI_predict_clearance_distribution
  ↓
ADMETAI_predict_toxicity

PATH 5: Clinical Trials

search_clinical_trials
  ↓
get_clinical_trial_conditions_and_interventions
  ↓
extract_clinical_trial_outcomes
  ↓
extract_clinical_trial_adverse_events

PATH 6: Post-Marketing Safety

FAERS_count_reactions_by_drug_event
  ↓
FAERS_count_seriousness_by_drug_event
  ↓
FAERS_count_outcomes_by_drug_event
  ↓
FAERS_count_death_related_by_drug
  ↓
FAERS_count_patient_age_distribution

PATH 7: Pharmacogenomics

PharmGKB_search_drugs
  ↓
PharmGKB_get_drug_details
  ↓
PharmGKB_get_clinical_annotations (for each gene)
  ↓
PharmGKB_get_dosing_guidelines

PATH 8: Literature

PubMed_search_articles (total count)
  ↓
PubMed_search_articles (recent papers)
  ↓
PubMed_search_articles (drug-focused)
  ↓
PubMed_get_cited_by (for key papers)

---

Fallback Chains

---

Rate Limiting & Best Practices

---

Input Format Requirements

SMILES

# ADMET-AI tools require list format
smiles = ["CC(=O)Oc1ccccc1C(=O)O"]  # List, not string

Drug Names (FAERS)

# Use uppercase, match FDA labeling
medicinalproduct = "METFORMIN"  # Not "metformin"

NCT IDs

# List format for batch queries
nct_ids = ["NCT01234567", "NCT02345678"]

ChEMBL IDs

# Full ID format
chembl_id = "CHEMBL1431"  # Not "1431"

PharmGKB IDs

# PA prefix for drugs
drug_id = "PA450657"  # Not "450657"