config.yml

---
# common parameters
n_cores: &cores 3

# This is an ordered dict that contains the order in which the steps are run.
run_sequence: !!omap
  - 1: 'basic_check'
  - 2: 'fill_data'
  - 3: 'infer_build'
  - 4: 'harmonize'
#  - 5: 'report_inflation_factors'
#  - 6: 'qq_manhattan_plots'
  - 7: 'write_pickle'
#  - 8: 'write_ldsc'
#  - 9: 'write_metal'
  - 10: 'write_regenie'
  - 11: "sort_alphabetically"


steps:
  basic_check: # see https://cloufield.github.io/gwaslab/Standardization/
    params:
      run: True
    gl_params:
      normalize: True
      n_cores: *cores
  infer_build: # see https://cloufield.github.io/gwaslab/InferBuild/
    params:
      run: True
  fill_data: # see https://cloufield.github.io/gwaslab/Conversion/#fill_data
    params:
      run: True
    gl_params:
      to_fill: ['MLOG10P', 'Z']
      overwrite: False
      extreme: False
  harmonize: # see https://cloufield.github.io/gwaslab/Harmonization/
    params:
      run: True
      preload_cache: True
      flip_allele: False
    gl_params:
      basic_check: False
      n_cores: *cores
      ref_seq: "data/GCA_000001405.14_GRCh37.p13_full_analysis_set.fna"
      ref_infer: "data/EUR.ALL.split_norm_af.1kgp3v5.hg19.vcf.gz"
      ref_alt_freq: "AF"
      remove: True
      removedup_args: {'mode': "d",'keep': 'first','keep_col': "MLOG10P" }
      inferstrand_args: {'cache_options': {'use_cache': True}} # if set to False but preload_cache is True, the (preloaded) cache will be used anyway
  liftover: # see https://cloufield.github.io/gwaslab/LiftOver/
    params:
      run: False
    gl_params:
      n_cores: *cores
      from_build: "19"
      to_build: "38"
      remove: True
  report_summary:
    params:
      run: False
  report_inflation_factors:
    params:
      run: True
      overwrite: False
      workspace: "if"
      workspace_subfolder: True
  report_min_pvalue:
    params:
      run: False
      workspace: "min_P"
      workspace_subfolder: True
      nrows: 20
  qq_manhattan_plots:
    params:
      run: True
      workspace: "plots"
      dist: 2
    gl_params:
      mode: 'mqq'
      stratified : True
      maf_bins: [[0, 0.01], [0.01, 0.05], [0.05, 1]]
      scaled: True
      build: "19"
      skip: 1
      sig_level: 5e-8
  cistrans_annotation:
    params:
      run: False
      workspace: "cistrans_tag"
    gl_params:
      target_gene_map_file_dir: "data/target_gene_mapping_file_v9-12-23.csv"
      map_file_chrom_col: "chromosome"
      map_file_start_col: "start"
      map_file_end_col: "end"
      map_file_gene_name: "Entrez_Gene_Name"
      gene_pos_file_dir: "data/glist-hg19.txt"
      genepos_file_genecol_name: "GENE_NAME"
      gwas_files_dir: "path/to/cojo/or/clump/files"
      prot_id_column_name: "SeqId"
      pval_column: "MLOG10P" # change with actual name of -log10p column
      snp_file_pos_col: "POS"
      snp_file_chrom_col: "#CHROM"
      pvalue_threshold: 7.30102 # -log10 of 5e-8
      region_buffer_value: 500000
      proximality_threshold: 5000
      columns_to_drop: []
  write_pickle:
    params:
      run: True
      overwrite: False
      workspace: "pickle"
      workspace_subfolder: False
  write_same_input_format:
    params:
      run: False
      workspace: "outputs"
      workspace_subfolder: True
    gl_params:
      to_csvargs:
        compression:
          method: "gzip"
          compresslevel: 1
          mtime: 1
  write_tsv:
    params:
      run: False
      workspace: "outputs"
      workspace_subfolder: True
    gl_params:
      fmt: "gwaslab"
  # see https://cloufield.github.io/gwaslab/Format/ for more details on writing options
  write_ldsc:
    params:
      run: False
      workspace: "ldsc"
      workspace_subfolder: True
    gl_params:
      fmt: "ldsc"
  write_metal:
    params:
      run: True
      workspace: "metal"
      workspace_subfolder: True
    gl_params:
      fmt: "metal"
  write_regenie:
    params:
      run: True
      workspace: "regenie"
      workspace_subfolder: True
    gl_params:
      fmt: "regenie"
  write_vcf:
    params:
      run: True
      workspace: "vcf"
      workspace_subfolder: True
      study_name_mask: [ True, True, True, False, False, False]
      study_name_sep: '.'
    gl_params:
      fmt: "vcf"
  sort_alphabetically:
    params:
      run: True
    gl_params:
      n_cores: *cores

# Filename transformation, e.g.: seq.3007.7.gwas.regenie.gz
filename_mask: [ True, True, True, False, False, False]
filename_sep: '.'

# IO
root_path: "results"
log_filename: "gwaspipe.log"
formatbook_path: " "     # fill this var if you want to provide a custom formatbook