@bgicli/bgicli 2.1.1 → 2.2.0

package/data/skills/tooluniverse-gwas-drug-discovery/SKILL.md

@@ -0,0 +1,576 @@
+---
+name: tooluniverse-gwas-drug-discovery
+description: Transform GWAS signals into actionable drug targets and repurposing opportunities. Performs locus-to-gene mapping, target druggability assessment, existing drug identification, safety profile evaluation, and clinical trial matching. Use when discovering drug targets from GWAS data, finding drug repurposing opportunities from genetic associations, or translating GWAS findings into therapeutic leads.
+---
+
+# GWAS-to-Drug Target Discovery
+
+Transform genome-wide association studies (GWAS) into actionable drug targets and repurposing opportunities.
+
+## Overview
+
+This skill bridges genetic discoveries from GWAS with drug development by:
+
+1. **Identifying genetic risk factors** - Finding genes associated with diseases
+2. **Assessing druggability** - Evaluating which genes can be targeted by drugs
+3. **Prioritizing targets** - Ranking candidates by genetic evidence strength
+4. **Finding existing drugs** - Discovering approved/investigational compounds
+5. **Identifying repurposing opportunities** - Matching drugs to new indications
+
+### Why This Matters
+
+**From Genetics to Therapeutics**: GWAS has identified thousands of disease-associated variants, but most haven't been translated into therapies. This skill accelerates that translation.
+
+**Success Stories**:
+- **PCSK9** (cholesterol) → Alirocumab, Evolocumab (approved 2015)
+- **IL-6R** (rheumatoid arthritis) → Tocilizumab (approved 2010)
+- **CTLA4** (autoimmunity) → Abatacept (approved 2005)
+- **CFTR** (cystic fibrosis) → Ivacaftor (approved 2012)
+
+**Genetic Evidence Doubles Success Rate**: Targets with genetic support have 2x higher probability of clinical approval (Nelson et al., Nature Genetics 2015).
+
+## Core Concepts
+
+### 1. GWAS Evidence Strength
+
+Not all genetic associations are equal. Consider:
+
+- **P-value** - Statistical significance (genome-wide: p < 5×10⁻⁸)
+- **Effect size (beta/OR)** - Magnitude of genetic effect
+- **Replication** - Confirmed in multiple studies
+- **Sample size** - Larger studies = more reliable
+- **Population diversity** - Validated across ancestries
+
+### 2. Druggability Criteria
+
+A good drug target must be:
+
+- **Accessible** - Protein location allows drug binding (extracellular > intracellular)
+- **Modality match** - Target class fits drug type (GPCR → small molecule, receptor → antibody)
+- **Tractable** - Binding pocket suitable for drug design
+- **Safe** - Minimal off-target effects, not essential in all tissues
+
+### 3. Target Prioritization Framework
+
+**GWAS Evidence (40%)**:
+- Multiple independent SNPs = stronger signal
+- Functional variants (missense > intronic)
+- Tissue-specific expression matches disease
+
+**Druggability (30%)**:
+- Known druggable protein family
+- Structural data available
+- Existing chemical matter
+
+**Clinical Evidence (20%)**:
+- Prior safety data
+- Validated disease models
+- Biomarker availability
+
+**Commercial Factors (10%)**:
+- Patent landscape
+- Market size
+- Competitive positioning
+
+### 4. Drug Repurposing Logic
+
+Repurposing works when:
+
+1. **Shared genetic architecture** - Same gene implicated in multiple diseases
+2. **Pathway overlap** - Related biological mechanisms
+3. **Opposite effects** - Drug's mechanism counteracts disease pathology
+4. **Proven safety** - Approved drug = de-risked
+
+**Example**: Metformin (T2D drug) being tested for:
+- Cancer (AMPK activation)
+- Aging (mitochondrial effects)
+- PCOS (insulin sensitization)
+
+## Workflow Steps
+
+### Step 1: GWAS Gene Discovery
+
+**Input**: Disease/trait name (e.g., "type 2 diabetes", "Alzheimer disease")
+
+**Process**:
+- Query GWAS Catalog for associations
+- Filter by significance threshold (p < 5×10⁻⁸)
+- Map variants to genes (nearest, eQTL, fine-mapping)
+- Aggregate evidence across studies
+
+**Output**: List of genes with genetic support
+
+**Tools Used**:
+- `gwas_get_associations_for_trait` - Get associations by disease
+- `gwas_search_associations` - Flexible search
+- `gwas_get_associations_for_snp` - SNP-specific associations
+- `OpenTargets_search_gwas_studies_by_disease` - Curated GWAS data
+- `OpenTargets_get_variant_credible_sets` - Fine-mapped loci with L2G predictions
+
+### Step 2: Druggability Assessment
+
+**Input**: Gene list from Step 1
+
+**Process**:
+- Check target class (GPCR, kinase, ion channel, etc.)
+- Assess tractability (antibody, small molecule)
+- Evaluate safety (expression profile, essentiality)
+- Check for tool compounds or crystal structures
+
+**Output**: Druggability score (0-1) + modality recommendations
+
+**Tools Used**:
+- `OpenTargets_get_target_tractability_by_ensemblID` - Druggability assessment
+- `OpenTargets_get_target_classes_by_ensemblID` - Target classification
+- `OpenTargets_get_target_safety_profile_by_ensemblID` - Safety data
+- `OpenTargets_get_target_genomic_location_by_ensemblID` - Genomic context
+
+### Step 3: Target Prioritization
+
+**Input**: Genes with GWAS + druggability data
+
+**Process**:
+- Calculate composite score: genetic evidence × druggability
+- Rank targets by score
+- Add qualitative factors (novelty, competitive landscape)
+- Generate target dossiers
+
+**Output**: Ranked list of drug target candidates
+
+**Scoring Formula**:
+```
+Target Score = (GWAS Score × 0.4) + (Druggability × 0.3) + (Clinical Evidence × 0.2) + (Novelty × 0.1)
+```
+
+### Step 4: Existing Drug Search
+
+**Input**: Prioritized target list
+
+**Process**:
+- Search drug-target associations (ChEMBL, DGIdb)
+- Find approved drugs, clinical candidates, tool compounds
+- Get mechanism of action, indication, phase
+- Check for off-label use or failed trials
+
+**Output**: Drug-target pairs with development status
+
+**Tools Used**:
+- `OpenTargets_get_associated_drugs_by_disease_efoId` - Known drugs for disease
+- `OpenTargets_get_drug_mechanisms_of_action_by_chemblId` - Drug MOA
+- `ChEMBL_get_target_activities` - Bioactivity data
+- `ChEMBL_get_drug_mechanisms` - Drug mechanisms
+- `ChEMBL_search_drugs` - Drug search
+
+### Step 5: Clinical Evidence
+
+**Input**: Drug candidates
+
+**Process**:
+- Check clinical trial history (ClinicalTrials.gov)
+- Review safety profile (FDA labels, adverse events)
+- Assess pharmacology (PK/PD, formulation)
+- Evaluate regulatory path
+
+**Output**: Clinical risk assessment
+
+**Tools Used**:
+- `FDA_get_adverse_reactions_by_drug_name` - Safety data
+- `FDA_get_active_ingredient_info_by_drug_name` - Drug composition
+- `OpenTargets_get_drug_warnings_by_chemblId` - Drug warnings
+
+### Step 6: Repurposing Opportunities
+
+**Input**: Approved drugs + new disease associations
+
+**Process**:
+- Match drug targets to new disease genes
+- Assess mechanistic fit (agonist vs antagonist)
+- Check contraindications
+- Estimate repurposing probability
+
+**Output**: Repurposing candidates with rationale
+
+**Repurposing Score**:
+- Genetic overlap: Gene targeted by drug = gene implicated in new disease
+- Clinical feasibility: Dosing, route, safety profile compatible
+- Regulatory path: Faster approval (Phase II vs Phase I)
+
+## Use Cases
+
+### Use Case 1: Novel Target Discovery for Rare Disease
+
+**Scenario**: Identify druggable targets for Huntington's disease
+
+**Steps**:
+1. Get GWAS hits for Huntington's → HTT, PDE10A, MSH3
+2. Assess druggability → PDE10A (phosphodiesterase) = high
+3. Find existing PDE10A inhibitors → Multiple tool compounds
+4. Recommendation: Develop selective PDE10A inhibitor
+
+**Clinical Context**:
+- HTT (huntingtin) = difficult to drug (large, scaffold protein)
+- PDE10A = modifier gene, GPCR-coupled, small molecule tractable
+- Precedent: PDE5 inhibitors (sildenafil) already approved
+
+### Use Case 2: Drug Repurposing for Common Disease
+
+**Scenario**: Find repurposing opportunities for Alzheimer's disease
+
+**Steps**:
+1. Get GWAS targets → APOE, CLU, CR1, PICALM, BIN1, TREM2
+2. Find drugs targeting these → Anti-inflammatory drugs (CR1, TREM2)
+3. Match approved drugs → Anakinra (IL-1R antagonist)
+4. Rationale: TREM2 links inflammation to neurodegeneration
+
+**Example Output**:
+```
+Repurposing Candidate: Anakinra
+- Target: IL-1R → affects TREM2 pathway
+- Current use: Rheumatoid arthritis (approved)
+- AD rationale: 3 GWAS genes in immune pathway
+- Clinical phase: Phase II trial in progress
+- Safety: Known profile, subcutaneous injection
+```
+
+### Use Case 3: Target Validation for Existing Drug Class
+
+**Scenario**: Validate new diabetes targets related to GLP-1 pathway
+
+**Steps**:
+1. Get T2D GWAS genes → TCF7L2, PPARG, KCNJ11, GLP1R
+2. GLP1R validated → Existing drug class (semaglutide, liraglutide)
+3. Check related genes → GIP, GIPR (glucose-dependent insulinotropic polypeptide)
+4. Outcome: Dual GLP-1/GIP agonists (tirzepatide, approved 2022)
+
+## Druggability Assessment Deep Dive
+
+### Target Classes (by Druggability)
+
+**Tier 1: High Druggability**
+- **GPCRs** (33% of approved drugs) - Extracellular binding, established chemistry
+- **Kinases** (18% of approved drugs) - ATP-competitive inhibitors, allosteric sites
+- **Ion channels** (15% of approved drugs) - Blocking/opening channels
+- **Nuclear receptors** - Ligand-binding domains
+
+**Tier 2: Moderate Druggability**
+- **Proteases** - Active site inhibitors
+- **Phosphatases** - Challenging selectivity
+- **Epigenetic targets** - Readers, writers, erasers
+
+**Tier 3: Difficult to Drug**
+- **Transcription factors** - No obvious binding pocket
+- **Scaffold proteins** - Large, flat surfaces
+- **RNA targets** - Emerging modality
+
+### Modality Selection
+
+**Small Molecules**:
+- Target: Intracellular proteins, enzymes
+- Advantages: Oral bioavailability, CNS penetration
+- Disadvantages: Off-target effects, development time
+- Examples: Kinase inhibitors, GPCR antagonists
+
+**Antibodies**:
+- Target: Extracellular proteins, receptors
+- Advantages: High specificity, long half-life
+- Disadvantages: Expensive, injection-only, no CNS
+- Examples: PD-1 inhibitors, TNF-α blockers
+
+**Antisense/RNAi**:
+- Target: mRNA (any gene)
+- Advantages: Sequence-specific, undruggable targets
+- Disadvantages: Delivery challenges, liver-centric
+- Examples: Patisiran (TTR), nusinersen (SMN)
+
+**Gene Therapy**:
+- Target: Genetic defects
+- Advantages: One-time treatment, curative potential
+- Disadvantages: Immunogenicity, manufacturing complexity
+- Examples: Luxturna (RPE65), Zolgensma (SMN1)
+
+## Clinical Translation Considerations
+
+### Regulatory Requirements
+
+**IND (Investigational New Drug) Application**:
+- Pharmacology and toxicology
+- Manufacturing information
+- Clinical protocols and investigator information
+
+**Clinical Trial Phases**:
+- **Phase I**: Safety, dosing (20-100 healthy volunteers)
+- **Phase II**: Efficacy, side effects (100-300 patients)
+- **Phase III**: Confirmatory trials (1,000-3,000 patients)
+- **Phase IV**: Post-market surveillance
+
+**Repurposing Advantages**:
+- Skip Phase I if dosing similar
+- Shorter timelines (2-4 years vs 10-15)
+- Lower costs ($50M vs $2B)
+
+### Success Rate Benchmarks
+
+**Traditional Drug Development** (Wong et al., Biostatistics 2019):
+- Phase I → II: 63%
+- Phase II → III: 31%
+- Phase III → Approval: 58%
+- Overall: 12% (from Phase I to approval)
+
+**With Genetic Evidence** (King et al., PLOS Genetics 2019):
+- Phase I → Approval: 24% (2× improvement)
+- Phase II → Approval: 38% vs 18% (no genetic support)
+
+### Cost and Timeline
+
+**Traditional Development**:
+- Pre-clinical: 3-6 years, $500M
+- Clinical trials: 6-7 years, $1-1.5B
+- Total: 10-15 years, $2-2.5B
+
+**Repurposing**:
+- Pre-clinical: 1-2 years, $50M
+- Clinical trials: 2-3 years, $100-200M
+- Total: 3-5 years, $150-250M
+
+## Best Practices
+
+### 1. Multi-Ancestry GWAS
+
+**Why**: Genetic architecture varies across populations
+
+**Approach**:
+- Include trans-ethnic meta-analyses
+- Check replication in multiple ancestries
+- Consider population-specific variants
+
+**Example**: APOL1 kidney disease variants (African ancestry-specific)
+
+### 2. Functional Validation
+
+**GWAS alone is not enough** - need mechanistic support:
+
+- **eQTL analysis**: Variant affects gene expression?
+- **pQTL analysis**: Variant affects protein levels?
+- **Colocalization**: GWAS + eQTL signals overlap?
+- **Fine-mapping**: Which variant(s) are causal?
+
+**Tools for validation**:
+- GTEx (tissue-specific expression)
+- ENCODE (regulatory elements)
+- gnomAD (variant frequency, constraint)
+
+### 3. Network and Pathway Analysis
+
+**Beyond Single Genes**:
+- Group GWAS hits by pathway (KEGG, Reactome)
+- Identify druggable nodes in disease network
+- Consider combination therapies
+
+**Example**: Alzheimer's GWAS →
+- Immune cluster (TREM2, CR1, CLU)
+- Lipid cluster (APOE, ABCA7)
+- Endocytosis (BIN1, PICALM)
+
+### 4. Safety Liability Assessment
+
+**Red Flags**:
+- Essential gene (loss-of-function lethal)
+- Broad expression (on-target toxicity)
+- Off-target kinase panel (promiscuity)
+- hERG inhibition (cardiotoxicity)
+- CYP450 interactions (drug-drug interactions)
+
+**Tools**:
+- gnomAD pLI (intolerance to loss-of-function)
+- GTEx expression (tissue specificity)
+- PharmaGKB (pharmacogenomics)
+
+### 5. Intellectual Property Landscape
+
+**Patent Considerations**:
+- Target patents (composition of matter)
+- Method of use patents (indication-specific)
+- Formulation patents (delivery)
+
+**Freedom to Operate**:
+- Existing patents on target
+- Blocking patents on drug class
+- Expired patents (generic opportunity)
+
+## Limitations and Caveats
+
+### GWAS Limitations
+
+**1. Association ≠ Causation**
+- Linkage disequilibrium = true causal variant may differ
+- Pleiotropy = gene affects multiple traits
+- Confounding = population stratification
+
+**Solution**: Fine-mapping, functional studies, Mendelian randomization
+
+**2. Missing Heritability**
+- Common variants explain ~10-50% of heritability
+- Rare variants, structural variants, epigenetics matter
+- Gene-environment interactions
+
+**Solution**: Whole-genome sequencing, family studies
+
+**3. Druggable ≠ Effective**
+- Can bind target ≠ modulates disease
+- Right direction (agonist vs antagonist)?
+- Right tissue (CNS penetration)?
+
+**Solution**: Experimental validation, disease models
+
+### Target Validation Challenges
+
+**1. Mouse Models ≠ Humans**
+- 95% of drugs work in mice, 5% in humans
+- Species differences (immune system)
+- Acute models ≠ chronic disease
+
+**Solution**: Human cell models (iPSCs, organoids), humanized mice
+
+**2. Genetic Perturbation ≠ Pharmacology**
+- Knockout = complete loss, drug = partial inhibition
+- Timing matters (developmental vs adult)
+- Compensation in knockout
+
+**Solution**: Inducible knockouts, tool compounds
+
+**3. Efficacy ≠ Safety**
+- On-target toxicity (essential gene)
+- Off-target effects (selectivity)
+- Dose-limiting side effects
+
+**Solution**: Therapeutic index assessment, biomarkers
+
+## Ethical and Regulatory Considerations
+
+### Human Genetics Research
+
+**Informed Consent**:
+- Secondary use of GWAS data
+- Return of results policies
+- Privacy protections (de-identification)
+
+**Equity**:
+- Most GWAS = European ancestry (78%)
+- Risk: Drugs may not work equally across populations
+- Solution: Diversify GWAS cohorts
+
+### Clinical Trials
+
+**Study Design**:
+- Stratification by genetics (precision medicine)
+- Adaptive trials (basket, umbrella designs)
+- Real-world evidence (pragmatic trials)
+
+**Patient Selection**:
+- Enrichment by genotype (higher response rate)
+- Ethics of genetic testing for trial entry
+- Cost-effectiveness of stratified medicine
+
+### Regulatory Pathways
+
+**FDA Breakthrough Therapy**:
+- Substantial improvement over existing
+- Expedited review (6 months vs 10 months)
+- Examples: CAR-T therapies, gene therapies
+
+**Accelerated Approval**:
+- Based on surrogate endpoints
+- Post-market confirmation required
+- Risk: Approval withdrawal if confirmatory fails
+
+## Resources and References
+
+### Databases
+
+**GWAS**:
+- [GWAS Catalog](https://www.ebi.ac.uk/gwas/) - Curated GWAS results
+- [Open Targets Genetics](https://genetics.opentargets.org/) - Fine-mapping, L2G
+- [PhenoScanner](http://www.phenoscanner.medschl.cam.ac.uk/) - Cross-trait lookups
+
+**Drugs**:
+- [ChEMBL](https://www.ebi.ac.uk/chembl/) - Bioactivity database
+- [DrugBank](https://go.drugbank.com/) - Comprehensive drug information
+- [DGIdb](https://www.dgidb.org/) - Drug-gene interactions
+
+**Targets**:
+- [Open Targets Platform](https://platform.opentargets.org/) - Target-disease associations
+- [PHAROS](https://pharos.nih.gov/) - Target development level (Tdark to Tclin)
+
+**Clinical**:
+- [ClinicalTrials.gov](https://clinicaltrials.gov/) - Clinical trial registry
+- [FDA Labels](https://labels.fda.gov/) - Drug labeling information
+
+### Key Literature
+
+**Genetic Evidence for Drug Targets**:
+- Nelson et al. (2015) *Nature Genetics* - Genetic support doubles clinical success
+- King et al. (2019) *PLOS Genetics* - Systematic analysis of target success
+
+**GWAS to Function**:
+- Visscher et al. (2017) *American Journal of Human Genetics* - 10 years of GWAS
+- Claussnitzer et al. (2020) *Nature Reviews Genetics* - From GWAS to biology
+
+**Drug Repurposing**:
+- Pushpakom et al. (2019) *Nature Reviews Drug Discovery* - Repurposing opportunities
+- Shameer et al. (2018) *Nature Biotechnology* - Computational repurposing
+
+**Success Stories**:
+- Plenge et al. (2013) *Nature Reviews Drug Discovery* - IL-6R to tocilizumab
+- Cohen et al. (2006) *Science* - PCSK9 to evolocumab
+
+## Disclaimer
+
+**For Research Purposes Only**
+
+This skill is designed for:
+- Target discovery and validation
+- Drug repurposing hypothesis generation
+- Preclinical research planning
+
+**NOT for**:
+- Clinical decision-making
+- Patient treatment recommendations
+- Regulatory submissions (without validation)
+
+**Important Notes**:
+- All targets require experimental validation
+- GWAS evidence is correlational, not causal
+- Regulatory approval requires extensive preclinical and clinical data
+- Consult domain experts (geneticists, pharmacologists, clinicians)
+
+**Liability**: The authors assume no liability for actions taken based on this analysis. All therapeutic development requires rigorous validation and regulatory oversight.
+
+## Version History
+
+- **v1.0.0** (2026-02-13): Initial release with GWAS-to-drug workflow
+  - Support for GWAS Catalog, Open Targets, ChEMBL, FDA tools
+  - Target discovery, druggability assessment, repurposing identification
+  - Comprehensive documentation with examples
+
+## Future Enhancements
+
+**Planned Features**:
+- Integration with UK Biobank for larger-scale GWAS
+- PheWAS (phenome-wide association studies) for pleiotropic effects
+- Mendelian randomization for causal inference
+- Network-based target prioritization
+- AI-powered structure-activity relationship (SAR) prediction
+- Clinical trial matching for repurposing candidates
+
+**Tool Additions**:
+- PDB (Protein Data Bank) for structural druggability
+- STRING for protein-protein interaction networks
+- DisGeNET for disease-gene associations
+- ClinVar for pathogenic variant interpretation
+
+## Contact
+
+For questions, issues, or contributions:
+- GitHub: [ToolUniverse Repository]
+- Documentation: [skills/tooluniverse-gwas-drug-discovery/]
+- Email: tooluniverse@example.com