ComprehensivePipeline

​

Overview

​

Constructor

ComprehensivePipeline(config: Config)

​

Attributes

• config (Config): Configuration object
• data_processor (DataProcessor): Data processing handler
• visualizer (Visualizer): Visualization handler
• exporter (ResultsExporter): Results export handler
• results (Dict[str, Any]): Dictionary storing pipeline results

​

Methods

​

run()

def run(data_path: str, output_dir: Optional[str] = None) -> Dict[str, Any]

• ImportError: If required dependencies (scanpy, pandas, numpy, matplotlib) are not available

1. Data Loading and Processing - Loads and preprocesses raw data using DataProcessor.process_from_raw()
2. Neighborhood Graph - Computes PCA (40 components) and k-nearest neighbors (k=15)
3. Dimensionality Reduction - Calculates UMAP for visualization
4. Cell Clustering - Performs Leiden clustering using the resolution from config
5. Communication Analysis - Calculates hub scores and communication patterns
6. Multi-Chamber Analysis - Identifies chamber-specific patterns
7. Comprehensive Visualization - Generates integrated dashboard with all visualizations
8. Report Generation - Creates comprehensive HTML/PDF report with all findings

​

save_results()

def save_results(output_dir: str) -> None

​

Usage Example

from heartmap.config import Config
from heartmap.pipelines import ComprehensivePipeline

# Create configuration with all parameters
config = Config(
    data_path="data/cardiac_tissue.h5ad",
    analysis={
        "resolution": 0.8,
        "n_neighbors": 15,
        "n_pcs": 40
    }
)

# Initialize comprehensive pipeline
pipeline = ComprehensivePipeline(config)

# Run complete analysis from raw data
results = pipeline.run(
    data_path="data/raw/cardiac_10x",
    output_dir="results/comprehensive"
)

# Access all results
adata = results['adata']

# Cell type annotations
cluster_labels = results['results']['annotation']['cluster_labels']
print(f"Identified {len(set(cluster_labels))} cell clusters")

# Communication analysis
hub_scores = results['results']['communication']['hub_scores']
print(f"Mean hub score: {hub_scores.mean():.3f}")

# Multi-chamber results
multi_chamber = results['results']['multi_chamber']
print(f"Multi-chamber analysis: {list(multi_chamber.keys())}")

# The AnnData object contains all annotations
print(f"\nadata.obs columns: {list(adata.obs.columns)}")
print(f"adata.obsm keys: {list(adata.obsm.keys())}")

​

Output Files

​

Data Files

• heartmap_complete.h5ad - Fully annotated AnnData object with all results
• Results exported via ResultsExporter

​

Visualizations

• figures/comprehensive_dashboard.png - Integrated multi-panel dashboard
• figures/umap_clusters.png - UMAP colored by cluster
• figures/qc_*.png - Quality control metrics
• figures/communication_*.png - Communication analysis plots
• figures/chamber_*.png - Multi-chamber analysis plots

​

Reports

• Comprehensive HTML/PDF report generated by ResultsExporter.generate_comprehensive_report()

​

Comparison with Other Pipelines

​

Best Practices

1. Use for complete analysis: The ComprehensivePipeline is ideal when you want all analyses in one workflow
2. Check output_dir size: Comprehensive analysis generates many files; ensure adequate disk space
3. Review configuration: All config parameters affect the comprehensive analysis
4. Inspect the dashboard: The comprehensive dashboard provides an overview of all results
5. Read the report: The generated report summarizes all findings with interpretations

​

Related Documentation