analysis.yaml

---
threads: -1  # Samtools & BWA (default: -1 = set dynamically based on available cores)
memory: -1   # Samtools (default: -1 = set dynamically based on free memory per core [MB])
tmpspace: 0  # Samtools (default: 0 [MB])

# I/O files
input:
  fasta: data/test.fasta  # filepath of ref. genome (haploid)
  seqids: [12, 22]  # zero or more SeqIDs (e.g. chromosomes)

output:
  basedir: data/out  # relative or absolute path
  genotype:   # diploid genomes
    - hmz     # homozygous
    - hmz-sv  # homozygous with SVs
    - htz-sv  # heterozygous with SVs

# registered I/O file extensions
filext:
  fasta: .fasta
  fasta_idx:
    - .fasta.ann  # BWA v0.6.x index files
    - .fasta.amb  #
    - .fasta.bwt  #
    - .fasta.pac  #
    - .fasta.sa   #
  fastq: .fq
  bam: .bam
  bam_idx: .bam.bai
  bed: .bed
  vcf: .vcf

simulation:
  # SURVIVOR parameters
  config: survivor.cfg
  svtype:
    dup: [0, 100, 10000]    # duplication: [count, min_len, max_len]
    inv: [0, 600, 800]      # inversion: ""
    tra: [10, 1000, 3000]   # translocation: ""
    indel: [10, 20, 500]    # insertion+deletion: ""
    invdel: [0, 600, 800]   # inversion+deletion: ""
    invdup: [0, 600, 800]   # inversion+duplication: ""
  # ART parameters
  seed: 1000
  profile: HSXt
  coverage: [10, 30]  # [cov1, cov2, ...]
  read:
    length: [150]  # [len1, len2, ...]
  insert:
    stdev: [10]  # standard deviation of the fragment length (bp)
    length: [500]  # [len1, len2, ...]