synthea-alldiseases/scripts/run_pipeline.sh

#!/bin/bash
# run_pipeline.sh
#
# This script runs the Nextflow pipeline for generating disease modules and synthetic patients

set -e  # Exit on error

# Default values
DISEASE_NAME=""
FORCE_GENERATE=false
GENERATE_PATIENTS=false
POPULATION=100
GENDER=0.5
MIN_AGE=0
MAX_AGE=90
SEED=""
ANALYZE_DATA=false
REPORT_FORMAT="html"
VERIFY_DOCKER=true

# Parse command line arguments
function show_help {
    echo "Usage: $0 [options] --disease \"Disease Name\""
    echo ""
    echo "Options:"
    echo "  --disease, -d NAME     Disease name to generate a module for (required)"
    echo "  --force-generate, -f   Force regeneration of disease module even if it exists"
    echo "  --patients, -p         Generate synthetic patients (default: false)"
    echo "  --population, -n NUM   Number of patients to generate (default: 100)"
    echo "  --gender, -g VALUE     Gender distribution (0-1 for % female, default: 0.5)"
    echo "  --min-age, -a NUM      Minimum patient age (default: 0)"
    echo "  --max-age, -m NUM      Maximum patient age (default: 90)"
    echo "  --seed, -s SEED        Random seed for reproducibility"
    echo "  --analyze, -A          Analyze patient data after generation"
    echo "  --report-format, -r FMT Report format for analysis (html, json, csv, default: html)"
    echo "  --skip-docker-check     Skip Docker container verification"
    echo "  --help, -h             Show this help message"
    echo ""
    echo "Example:"
    echo "  $0 --disease \"Multiple Sclerosis\" --patients --population 50"
    exit 1
}

# Parse arguments
while [[ $# -gt 0 ]]; do
    case "$1" in
        --disease|-d)
            DISEASE_NAME="$2"
            shift 2
            ;;
        --force-generate|-f)
            FORCE_GENERATE=true
            shift
            ;;
        --patients|-p)
            GENERATE_PATIENTS=true
            shift
            ;;
        --population|-n)
            POPULATION="$2"
            shift 2
            ;;
        --gender|-g)
            GENDER="$2"
            shift 2
            ;;
        --min-age|-a)
            MIN_AGE="$2"
            shift 2
            ;;
        --max-age|-m)
            MAX_AGE="$2"
            shift 2
            ;;
        --seed|-s)
            SEED="$2"
            shift 2
            ;;
        --analyze|-A)
            ANALYZE_DATA=true
            shift
            ;;
        --report-format|-r)
            REPORT_FORMAT="$2"
            shift 2
            ;;
        --skip-docker-check)
            VERIFY_DOCKER=false
            shift
            ;;
        --help|-h)
            show_help
            ;;
        *)
            echo "Unknown option: $1"
            show_help
            ;;
    esac
done

# Check if disease name is provided
if [ -z "$DISEASE_NAME" ]; then
    echo "ERROR: Disease name is required!"
    show_help
fi

# Check if Docker container is running if we need to generate patients
if [ "$VERIFY_DOCKER" = true ] && [ "$GENERATE_PATIENTS" = true ]; then
    CONTAINER_RUNNING=$(docker ps | grep synthea | wc -l)
    if [ "$CONTAINER_RUNNING" -eq 0 ]; then
        echo "No Synthea Docker containers are running!"
        echo "Running environment setup script to start containers..."
        
        if [ -f "./scripts/prepare_environment.sh" ]; then
            ./scripts/prepare_environment.sh
        elif [ -f "./prepare_environment.sh" ]; then
            ./prepare_environment.sh
        else
            echo "ERROR: prepare_environment.sh not found!"
            exit 1
        fi
    else
        echo "Synthea Docker container is running. Proceeding with pipeline execution."
    fi
fi

# Create params.json file with our configuration
echo "Creating params.json file with pipeline configuration..."
cat > params.json << EOF
{
  "disease_name": "$DISEASE_NAME",
  "modules_dir": "modules",
  "output_dir": "output",
  "generate_patients": $GENERATE_PATIENTS,
  "population": $POPULATION,
  "gender": $GENDER,
  "min_age": $MIN_AGE,
  "max_age": $MAX_AGE,
  "analyze_patient_data": $ANALYZE_DATA,
  "report_format": "$REPORT_FORMAT",
  "force_generate": $FORCE_GENERATE,
  "publish_dir": "published_output"
EOF

# Add seed if provided
if [ ! -z "$SEED" ]; then
    echo ",  \"seed\": $SEED" >> params.json
fi

# Close JSON object
echo "}" >> params.json

# Display execution details
echo "=================================================="
echo "Running Synthea Pipeline for: $DISEASE_NAME"
echo "Generate patients: $GENERATE_PATIENTS"
if [ "$GENERATE_PATIENTS" = true ]; then
    echo "Population: $POPULATION"
    echo "Gender ratio (proportion female): $GENDER"
    echo "Age range: $MIN_AGE-$MAX_AGE"
    if [ ! -z "$SEED" ]; then
        echo "Random seed: $SEED"
    fi
fi
echo "Force module generation: $FORCE_GENERATE"
echo "Analyze patient data: $ANALYZE_DATA"
echo "=================================================="

# Execute the Nextflow command
echo "Starting Nextflow pipeline..."
nextflow run main.nf

echo ""
echo "Pipeline execution complete!"
if [ "$GENERATE_PATIENTS" = true ]; then
    echo "Check the 'output' directory and 'published_output/$DISEASE_NAME_NORMALIZED' for generated patient data."
    if [ "$ANALYZE_DATA" = true ]; then
        echo "Analysis reports can be found in 'published_output/$DISEASE_NAME_NORMALIZED/analysis'."
    fi
fi
echo "Generated modules can be found in the 'modules' directory and 'published_output/modules'."
echo ""