129 lines
4.4 KiB
Plaintext
129 lines
4.4 KiB
Plaintext
#!/usr/bin/env nextflow
|
|
|
|
nextflow.enable.dsl=2
|
|
|
|
/*
|
|
* Synthea Disease Module Generator Pipeline
|
|
*
|
|
* A Nextflow pipeline to generate and manage Synthea disease modules
|
|
*/
|
|
|
|
// Load API key from .env file if it exists
|
|
def envFile = file('.env')
|
|
if (envFile.exists()) {
|
|
envFile.eachLine { line ->
|
|
def (key, value) = line.tokenize('=')
|
|
if (key && value && key.trim() == 'ANTHROPIC_API_KEY') {
|
|
params.anthropic_api_key = value.trim()
|
|
}
|
|
}
|
|
}
|
|
|
|
// Default parameters
|
|
params.disease_name = null // Disease name to generate patients for
|
|
params.output_dir = "/mnt/OmicNAS/private/old/olamide/synthea/output/new" // Output directory
|
|
params.modules_dir = "src/main/resources/modules" // Directory for module files
|
|
params.population = 100 // Number of patients to generate
|
|
params.gender = 0.5 // Decimal representing proportion female (0.0-1.0)
|
|
params.min_age = 0 // Minimum age of generated patients
|
|
params.max_age = 90 // Maximum age of generated patients
|
|
params.seed = null // Random seed for reproducibility
|
|
params.help = false // Show help message
|
|
|
|
// Show help message
|
|
if (params.help) {
|
|
log.info """
|
|
Synthea Patient Generator
|
|
========================
|
|
Usage: nextflow run main.nf --disease_name "Disease Name"
|
|
|
|
Required Arguments:
|
|
--disease_name Disease name to generate patients for
|
|
|
|
Optional Arguments:
|
|
--modules_dir Module directory (default: modules)
|
|
--output_dir Output directory (default: output)
|
|
--population Number of patients (default: 100)
|
|
--gender Gender ratio - female proportion 0.0-1.0 (default: 0.5)
|
|
--min_age Minimum age (default: 0)
|
|
--max_age Maximum age (default: 90)
|
|
--seed Random seed (default: random)
|
|
"""
|
|
exit 0
|
|
}
|
|
|
|
// Validate required parameters
|
|
if (!params.disease_name && !params.help) {
|
|
error "Disease name is required. Please specify with --disease_name"
|
|
}
|
|
|
|
// Process to check if module exists and generate it if needed
|
|
process checkAndGetModule {
|
|
container 'synthea-module-generator'
|
|
publishDir "${params.modules_dir}", mode: 'copy'
|
|
|
|
input:
|
|
val diseaseName
|
|
|
|
output:
|
|
path "*.json", emit: module_file
|
|
|
|
script:
|
|
def moduleFilename = diseaseName.toLowerCase().replaceAll(' ', '_') + '.json'
|
|
def fullPath = "/app/src/main/resources/modules/${moduleFilename}"
|
|
"""
|
|
echo "Looking for module at ${fullPath}"
|
|
if [ -f "${fullPath}" ]; then
|
|
echo "Module exists, copying..."
|
|
cp "${fullPath}" .
|
|
else
|
|
echo "Module not found, generating..."
|
|
python3 /app/module_generator/module_generator.py --disease "${diseaseName}" --output "${moduleFilename}"
|
|
if [ -f "${moduleFilename}" ]; then
|
|
echo "Successfully generated module"
|
|
else
|
|
echo "Error: Failed to generate module"
|
|
exit 1
|
|
fi
|
|
fi
|
|
"""
|
|
}
|
|
|
|
// Process to generate synthetic patients
|
|
process generatePatients {
|
|
container 'synthea-module-generator'
|
|
publishDir "${params.output_dir}/${diseaseName.toLowerCase().replaceAll(' ', '_')}", mode: 'copy'
|
|
|
|
input:
|
|
val diseaseName
|
|
path moduleFile
|
|
|
|
output:
|
|
path "m/*", optional: true
|
|
path "f/*", optional: true
|
|
|
|
script:
|
|
def moduleBasename = diseaseName.toLowerCase().replaceAll(' ', '_')
|
|
def genderArg = params.gender < 0.5 ? "M" : (params.gender > 0.5 ? "F" : "B")
|
|
def seedValue = params.seed ?: new Random().nextInt(1000000)
|
|
"""
|
|
# Copy module and run Synthea
|
|
cp "${moduleFile}" /app/modules/
|
|
cd /app && ./run_synthea -p ${params.population} -g ${genderArg} -m ${moduleBasename} -a ${params.min_age}-${params.max_age} -s ${seedValue}
|
|
|
|
# Organize output by gender
|
|
mkdir -p m f
|
|
find /app/output/fhir -type f -name "*.json" ! -name "*hospital*" ! -name "*practitioner*" | xargs grep -l '"gender":"male"' | xargs -I{} cp {} m/
|
|
find /app/output/fhir -type f -name "*.json" ! -name "*hospital*" ! -name "*practitioner*" | xargs grep -l '"gender":"female"' | xargs -I{} cp {} f/
|
|
"""
|
|
}
|
|
|
|
// Define workflow
|
|
workflow {
|
|
// First check if the module exists
|
|
checkAndGetModule(params.disease_name)
|
|
|
|
// Then generate patients
|
|
generatePatients(params.disease_name, checkAndGetModule.out.module_file)
|
|
}
|