Skip to content

Bioinformatics Flow

The Bioinformatics flow provides comprehensive multi-source data fusion and integrative reasoning capabilities for biological research questions.

Overview

The Bioinformatics flow implements a sophisticated data fusion pipeline that integrates multiple biological databases and applies advanced reasoning to provide comprehensive answers to complex biological questions.

Architecture

graph TD
    A[Research Query] --> B[Parse Stage]
    B --> C[Intent Classification]
    C --> D[Data Source Selection]
    D --> E[Fuse Stage]
    E --> F[Data Integration]
    F --> G[Quality Assessment]
    G --> H[Reason Stage]
    H --> I[Evidence Integration]
    I --> J[Cross-Validation]
    J --> K[Final Reasoning]
    K --> L[Comprehensive Report]

Configuration

Basic Configuration

# Enable bioinformatics flow
flows:
  bioinformatics:
    enabled: true
    data_sources:
      go:
        enabled: true
        evidence_codes: ["IDA", "EXP"]
      pubmed:
        enabled: true
        max_results: 50

Advanced Configuration

# configs/statemachines/flows/bioinformatics.yaml
enabled: true

data_sources:
  go:
    enabled: true
    api_base_url: "https://api.geneontology.org"
    evidence_codes: ["IDA", "EXP", "TAS", "IMP"]
    year_min: 2020
    quality_threshold: 0.85
    max_annotations: 1000

  pubmed:
    enabled: true
    api_base_url: "https://eutils.ncbi.nlm.nih.gov/entrez/eutils"
    max_results: 100
    include_abstracts: true
    year_min: 2020
    relevance_threshold: 0.7

  geo:
    enabled: false
    max_datasets: 10
    sample_threshold: 50

  cmap:
    enabled: false
    max_profiles: 100
    correlation_threshold: 0.8

fusion:
  quality_threshold: 0.8
  max_entities: 1000
  cross_reference_enabled: true
  evidence_prioritization: true

reasoning:
  model: "anthropic:claude-sonnet-4-0"
  confidence_threshold: 0.75
  integrative_approach: true
  evidence_codes_priority: ["IDA", "EXP", "TAS", "IMP", "IGI"]

Data Sources

Gene Ontology (GO)

# GO annotation retrieval
go_annotations = await go_tool.query_annotations(
    gene_id="TP53",
    evidence_codes=["IDA", "EXP"],
    year_min=2020,
    max_results=100
)

# Process annotations
for annotation in go_annotations:
    print(f"GO Term: {annotation.go_id}")
    print(f"Evidence: {annotation.evidence_code}")
    print(f"Reference: {annotation.reference}")

PubMed Integration

# Literature search and retrieval
pubmed_results = await pubmed_tool.search_and_fetch(
    query="TP53 AND cancer AND apoptosis",
    max_results=50,
    include_abstracts=True,
    year_min=2020
)

# Extract key information
for paper in pubmed_results:
    print(f"PMID: {paper.pmid}")
    print(f"Title: {paper.title}")
    print(f"Abstract: {paper.abstract[:200]}...")

Expression Data (GEO)

# Gene expression analysis
geo_datasets = await geo_tool.query_datasets(
    gene_symbol="TP53",
    conditions=["cancer", "normal"],
    sample_count_min=50
)

# Analyze differential expression
for dataset in geo_datasets:
    print(f"Dataset: {dataset.accession}")
    print(f"Expression fold change: {dataset.fold_change}")

Perturbation Data (CMAP)

# Drug perturbation analysis
cmap_profiles = await cmap_tool.query_perturbations(
    gene_targets=["TP53"],
    compounds=["doxorubicin", "cisplatin"],
    correlation_threshold=0.8
)

# Identify drug-gene relationships
for profile in cmap_profiles:
    print(f"Compound: {profile.compound}")
    print(f"Correlation: {profile.correlation}")

Usage Examples

Gene Function Analysis

uv run deepresearch \
  flows.bioinformatics.enabled=true \
  question="What is the function of TP53 gene based on GO annotations and recent literature?"

Drug-Target Analysis

uv run deepresearch \
  flows.bioinformatics.enabled=true \
  question="Analyze the relationship between drug X and protein Y using expression profiles and interactions"

Multi-Source Integration

uv run deepresearch \
  flows.bioinformatics.enabled=true \
  question="What is the likely function of protein P12345 based on its structure, expression, and GO annotations?"

Evidence Integration

The bioinformatics flow uses sophisticated evidence integration:

Evidence Code Prioritization

  1. IDA (Inferred from Direct Assay) - Gold standard experimental evidence
  2. EXP (Inferred from Experiment) - Experimental evidence
  3. TAS (Traceable Author Statement) - Curated expert knowledge
  4. IMP (Inferred from Mutant Phenotype) - Genetic evidence
  5. IGI (Inferred from Genetic Interaction) - Interaction evidence

Cross-Database Validation

  • Consistency checks across GO, UniProt, and literature
  • Temporal relevance validation (recent vs. outdated annotations)
  • Species-specific annotation filtering
  • Confidence score aggregation

Quality Assessment

Data Quality Metrics

# Quality assessment framework
quality_metrics = {
    "annotation_quality": 0.85,  # GO annotation confidence
    "literature_relevance": 0.92,  # PubMed result relevance
    "expression_consistency": 0.78,  # GEO data consistency
    "cross_reference_agreement": 0.89  # Agreement across sources
}

# Overall quality score
overall_quality = sum(quality_metrics.values()) / len(quality_metrics)

Confidence Scoring

  • High Confidence (>0.85): Strong evidence from multiple sources
  • Medium Confidence (0.7-0.85): Good evidence with some inconsistencies
  • Low Confidence (<0.7): Weak or conflicting evidence

Reasoning Integration

Integrative Reasoning Process

# Multi-source evidence synthesis
reasoning_result = await reasoning_agent.integrate_evidence(
    go_annotations=go_data,
    literature=pubmed_data,
    expression=geo_data,
    interactions=string_data,
    confidence_threshold=0.75
)

# Generate comprehensive explanation
final_answer = {
    "primary_function": reasoning_result.primary_function,
    "evidence_summary": reasoning_result.evidence_summary,
    "confidence_score": reasoning_result.confidence_score,
    "alternative_functions": reasoning_result.alternative_functions,
    "research_gaps": reasoning_result.research_gaps
}

Output Formats

Structured Results

{
  "query": "TP53 gene function analysis",
  "data_sources_used": ["go", "pubmed", "geo"],
  "results": {
    "primary_function": "Tumor suppressor protein",
    "molecular_function": ["DNA binding", "Transcription regulation"],
    "biological_process": ["Cell cycle arrest", "Apoptosis"],
    "cellular_component": ["Nucleus", "Cytoplasm"],
    "evidence_strength": "high",
    "confidence_score": 0.89
  },
  "literature_summary": {
    "total_papers": 1247,
    "relevant_papers": 89,
    "key_findings": ["TP53 mutations in 50% of cancers"],
    "recent_trends": ["Immunotherapy targeting"]
  }
}

Visualization Outputs

  • GO term enrichment plots
  • Expression heatmap visualizations
  • Protein interaction networks
  • Literature co-occurrence graphs

Integration Examples

With PRIME Flow

uv run deepresearch \
  flows.prime.enabled=true \
  flows.bioinformatics.enabled=true \
  question="Design TP53-targeted therapy based on functional annotations and interaction data"

With DeepSearch Flow

uv run deepresearch \
  flows.bioinformatics.enabled=true \
  flows.deepsearch.enabled=true \
  question="Latest research on TP53 mutations and their therapeutic implications"

Advanced Features

Custom Data Sources

# Add custom data source
custom_source = {
    "name": "my_database",
    "type": "protein_interactions",
    "api_endpoint": "https://my-api.com",
    "authentication": {"token": "my-token"}
}

# Register for use
config_manager.add_data_source(custom_source)

Evidence Weighting

# Customize evidence weighting
evidence_weights = {
    "IDA": 1.0,    # Direct experimental evidence
    "EXP": 0.9,    # Experimental evidence
    "TAS": 0.8,    # Curated expert knowledge
    "IMP": 0.7,    # Genetic evidence
    "IGI": 0.6     # Interaction evidence
}

# Apply to reasoning
reasoning_config.evidence_weights = evidence_weights

Best Practices

  1. Multi-Source Validation: Always use multiple data sources for validation
  2. Evidence Prioritization: Focus on high-quality, recent evidence
  3. Cross-Reference Checking: Validate findings across different databases
  4. Temporal Filtering: Consider recency of annotations and literature
  5. Species Consideration: Account for species-specific differences

Troubleshooting

Common Issues

Low-Quality Results: 

# Increase quality thresholds
flows.bioinformatics.fusion.quality_threshold=0.85
flows.bioinformatics.data_sources.go.quality_threshold=0.9

Slow Data Retrieval: 

# Optimize data source settings
flows.bioinformatics.data_sources.pubmed.max_results=50
flows.bioinformatics.data_sources.go.max_annotations=500

Integration Failures: 

# Enable cross-reference validation
flows.bioinformatics.fusion.cross_reference_enabled=true
flows.bioinformatics.reasoning.integrative_approach=true

For more detailed information, see the Tool Development Guide and Data Types API Reference.