evaluation_pipeline.yml

io:
  path_mdata: /cellar/users/aklie/opt/gene_program_evaluation/examples/inference/iPSC_EC/cNMF/cNMF_30_0.2_gene_names.h5mu
  path_out: /cellar/users/aklie/opt/gene_program_evaluation/examples/evaluation/iPSC_EC/cNMF/cNMF_30
  data_key: rna
  prog_key: cNMF

categorical_association:
  categorical_keys: 
    - sample
  pseudobulk_key: null
  test: pearsonr
  mode: one_vs_all
  n_jobs: -1
  inplace: false

perturbation_association:
  groupby_key: sample
  collapse_targets: true
  pseudobulk: false
  reference_targets:
  - non-targeting
  n_jobs: -1
  inplace: false

gene_set_enrichment:
  prog_nam: null
  organism: 'human'
  libraries:
    - 'Reactome_2022'
    - 'GO_Biological_Process_2023'
  method: 'fisher'
  database: 'enrichr'
  n_top: 500
  low_cutoff: 0.0
  n_jobs: -1
  inplace: false
  user_geneset: null
  max_size: 500
  min_size: 5

trait_enrichment:
  gwas_data: /cellar/users/aklie/opt/gene_program_evaluation/smk/resources/OpenTargets_L2G_Filtered.csv.gz
  prog_nam: null
  library: OT_GWAS
  n_jobs: -1
  inplace: false
  key_column: trait_efos
  gene_column: gene_name
  method: fisher
  n_top: 500
  low_cutoff: 0.0
  min_size: 5
  max_size: 2000
  metadata: /cellar/users/aklie/opt/gene_program_evaluation/smk/resources/OpenTargets_L2G_Filtered.csv.gz
  pval_col: adj_pval
  enrich_geneset_id_col: term
  metadata_geneset_id_col: trait_efos
  color_category_col: trait_category
  program_name_col: program_name
  annotation_cols:
  - enrichment
  - trait_reported
  - genes
  - study_id
  - pmid

motif_enrichment:
  groupby_key: sample
  motif_file: /cellar/users/aklie/opt/gene_program_evaluation/src/tests/test_data/motifs.meme
  seq_file: /cellar/users/aklie/data/ref/genomes/hg38/hg38.fa
  loci_files: 
    - /cellar/users/aklie/opt/gene_program_evaluation/examples/datasets/iPSC_EC/EnhancerPredictions_D0_2024_09_07.txt
    - /cellar/users/aklie/opt/gene_program_evaluation/examples/datasets/iPSC_EC/EnhancerPredictions_sample_D1_2024_09_07.txt
    - /cellar/users/aklie/opt/gene_program_evaluation/examples/datasets/iPSC_EC/EnhancerPredictions_sample_D2_2024_09_07.txt
    - /cellar/users/aklie/opt/gene_program_evaluation/examples/datasets/iPSC_EC/EnhancerPredictions_sample_D3_2024_09_07.txt
  names:
    - 'D0'
    - 'sample_D1'
    - 'sample_D2'
    - 'sample_D3'
  output_loc: null
  window: 1000
  threshold: 0.001
  eps: 0.001
  reverse_complement: true
  sig: 0.05
  num_genes: null
  correlation: pearsonr
  n_jobs: -1
  inplace: false

explained_variance:
  layer: X
  n_jobs: 1
  inplace: false