config.yml

# Cancer@Home Configuration

app:
  name: "Cancer@Home v2"
  version: "2.0.0"
  host: "localhost"
  port: 5000
  debug: true

neo4j:
  uri: "bolt://localhost:7687"
  username: "neo4j"
  password: "cancer123"
  database: "neo4j"
  max_connection_lifetime: 3600
  max_connection_pool_size: 50

gdc:
  api_url: "https://api.gdc.cancer.gov"
  data_endpoint: "/data"
  files_endpoint: "/files"
  cases_endpoint: "/cases"
  download_dir: "./data/gdc"
  max_retries: 3
  timeout: 300

boinc:
  project_url: "http://localhost:8000"  # Local BOINC server
  username: "cancer_volunteer"
  password: "volunteer123"
  work_dir: "./data/boinc"
  max_concurrent_tasks: 4

pipeline:
  fastq:
    quality_threshold: 20
    min_length: 50
    output_dir: "./data/processed/fastq"
  
  blast:
    database: "nt"
    evalue: 0.001
    num_threads: 4
    output_dir: "./data/processed/blast"
  
  variant_calling:
    min_coverage: 10
    min_allele_frequency: 0.05
    output_dir: "./data/processed/variants"

data:
  cache_dir: "./data/cache"
  max_cache_size_gb: 10
  projects:
    - "TCGA-BRCA"  # Breast Cancer
    - "TCGA-LUAD"  # Lung Adenocarcinoma
    - "TCGA-COAD"  # Colon Adenocarcinoma
    - "TCGA-GBM"   # Glioblastoma
    - "TARGET-AML" # Acute Myeloid Leukemia

logging:
  level: "INFO"
  format: "%(asctime)s - %(name)s - %(levelname)s - %(message)s"
  file: "./logs/cancer_at_home.log"
  max_bytes: 10485760  # 10MB
  backup_count: 5