example.config.yml

# Example config for scCoAnnotate annotation and benchmarking workflow 

# target directory 
output_dir: <output directory for the annotation workflow>

### Description of some non-tool specific parameters 
references:
      <reference_name>:
            expression: <path to counts>
            labels: <path to labels>
            # Convert gene symbols in reference from mouse to human
            # Accepted values: True, False
            output_dir_benchmark: <output directory for the benchmarking workflow>
            convert_ref_mm_to_hg: False
            # The ontology permits to specify different level of labels granularity.
            # These parameters are optional - delete if unused
            ontology:
                  # Path to the csv containing the ontology path. Each column represents a different
                  # granularity of labels. The columns should be named.
                  ontology_path: <path to ontology.csv>
                  # The column name(s) of the granularity to use, from the ontology file.
                  # This parameter can take multiple column names, therefore they should be put in a list
                  # (ex: ['level']     ['level1', 'level2'])
                  ontology_column: <ontology_column to use>
            # Some references are too big and cannot be used efficiently
            # the following parameters permit to downsample the reference
            downsample:
                  # The number of cells to downsample to
                  # If the value is > 1, it specifies the number of cells to select (ex: 500 will select 500 cells)
                  # If the value is < 1, it is interpreted as a fraction of cells to keep (ex: 0.25 will select 25% of the cells)
                  value: 500
                  # Should the sample keep the same stratification as the complete dataset?
                  # Accepted values: True, False
                  stratified: True
            # The minimal number of cells that each cluster should have, in the reference
            # Clusters with less cells will be filtered out from the reference
            min_cells_per_cluster: 100

# path to query datasets (cell x gene raw counts)
query_datasets:
      <query_name_1>: <path to counts 1>
      <query_name_2>: <path to counts 2>
      <query_name_3>: <path to counts 3>

# classifiers to run
tools_to_run:
      - tool1
      - tool2

# consensus method
consensus:
        tools: 
            - 'all'
        type:
            majority:
                  # (ex: [2]     [2,3,4])
                  min_agree: <minimum agreemeent to use>
            CAWPE:
                  #(ex: ['CAWPE_T'], ['CAWPE_T','CAWPE_CT'])
                  mode: <CAWPE MODE>
                  #(ex: [4], [2,3,4])
                  alpha: <alpha value>

# benchmark parameters 
benchmark:
      n_folds: <number of folds to use in the benchmarking>