xrefs-external.template

include "xrefs-common.conseq"

# SET_TAIGA_PREPROCESSOR virtual_dataset_id "public-24q4-31a4.61"

let RELEASE_LABEL="Public 24Q4"

add-if-missing {
    "type": "depmap_data_taiga_id",
    "dataset_id": PREPROCESS_TAIGA_ID(virtual_dataset_id)
}

# Sample info file
add-if-missing {
  "type": "sample_info_dataset_id",
  "dataset_id": PREPROCESS_TAIGA_ID(virtual_dataset_id, "Model"),
}

add-if-missing {
    "type": "crispr-confounder-parameters",
    "achilles_qc_report_taiga_id":  PREPROCESS_TAIGA_ID(virtual_dataset_id, "AchillesScreenQCReport"),
    "crispr_screen_map_taiga_id": PREPROCESS_TAIGA_ID(virtual_dataset_id, "CRISPRScreenMap")
}

# Chronos Combined
add-if-missing {
  "type": "raw-dep-prob-matrix",
  "dataset_id": PREPROCESS_TAIGA_ID(virtual_dataset_id, "CRISPRGeneDependency"),
  "label": "Chronos_Combined",
  "rows": "cell-lines"
}

add-if-missing {
  "type": "crispr-screen-sequence-map",
  "dataset_id": PREPROCESS_TAIGA_ID(virtual_dataset_id, "ScreenSequenceMap")
}


add-if-missing {
  "type": "raw-dep-matrix",
  "dataset_id": PREPROCESS_TAIGA_ID(virtual_dataset_id, "CRISPRGeneEffect"),
  "label": "Chronos_Combined",
  "rows": "cell-lines",
  "confounders_label": "crispr-confounders",
}

# Chronos Achilles
add-if-missing {
  "type": "confounders-matrix-essential-genes",
  "dataset_id": PREPROCESS_TAIGA_ID(virtual_dataset_id, "AchillesCommonEssentialControls")
}

add-if-missing {
  "type": "confounders-matrix-nonessential-genes",
  "dataset_id": PREPROCESS_TAIGA_ID(virtual_dataset_id, "AchillesNonessentialControls")
}

# Mutation table
add-if-missing {
  "type": "mutation-maf",
  "dataset_id": PREPROCESS_TAIGA_ID(virtual_dataset_id, "OmicsSomaticMutations")
}

add-if-missing {
    "type": "other-taiga-dataset",
    "category": "fusions",
    "dataset_id": PREPROCESS_TAIGA_ID(virtual_dataset_id, "OmicsFusionFiltered")
}

# CRISPRInferredCommonEssentials
add-if-missing {
  "type": "crispr-inferred-common-essentials",
  "dataset_id": PREPROCESS_TAIGA_ID(virtual_dataset_id, "CRISPRInferredCommonEssentials")
}

# biomarker-matrix
add-if-missing {
  'type': 'raw-expr-matrix',
  'dataset_id' : PREPROCESS_TAIGA_ID(virtual_dataset_id, "OmicsExpressionProteinCodingGenesTPMLogp1"),
  'category' : 'expression',
}

# Log2-transformed CNGene data
add-if-missing {
  'type': 'biomarker-needing-transpose',
  'dataset_id' : PREPROCESS_TAIGA_ID(virtual_dataset_id, "PortalOmicsCNGeneLog2"),
  'category': 'copy-number-relative'
}


add-if-missing {
  "type": "raw-mutations-bool-matrix",
  "dataset_id": PREPROCESS_TAIGA_ID(virtual_dataset_id, "OmicsSomaticMutationsMatrixDamaging"),
  "category": "damaging"
}

add-if-missing {
  "type": "raw-mutations-bool-matrix",
  "dataset_id": PREPROCESS_TAIGA_ID(virtual_dataset_id, "OmicsSomaticMutationsMatrixHotspot"),
  "category": "hotspot"
}


# These below are used to label datasets. It's a little odd that the display label is stored 
# seperately from the artifact with the data (and so we'll have to update both when versions change)
# but doing it this way saves us from having to re-run rules when the display name changes

add-if-missing {
    'type': 'dataset-display-name',
    'display_name': 'Copy Number {{ config.RELEASE_LABEL }} (Log2 transformed)',
    'label': 'copy_number_relative',
    'dataset_id': PREPROCESS_TAIGA_ID(virtual_dataset_id, "PortalOmicsCNGeneLog2")
}

add-if-missing {
    'type': 'dataset-display-name',
    'display_name': 'Expression {{ config.RELEASE_LABEL }}',
    'label': 'expression',
    'dataset_id' : PREPROCESS_TAIGA_ID(virtual_dataset_id, "OmicsExpressionProteinCodingGenesTPMLogp1"),
}

add-if-missing {
    'type': 'dataset-display-name',
    'display_name': 'Fusions {{ config.RELEASE_LABEL }}',
    'label': 'fusions',
    'dataset_id' : PREPROCESS_TAIGA_ID(virtual_dataset_id, "OmicsFusionFiltered"),
}

add-if-missing {
    'type': 'dataset-display-name',
    'display_name': 'Mutation {{ config.RELEASE_LABEL }}',
    'label': 'mutation_pearson',
    'dataset_id' : PREPROCESS_TAIGA_ID(virtual_dataset_id, "OmicsSomaticMutations"),
}

add-if-missing {
    'type': 'dataset-display-name',
    'display_name': 'CRISPR (DepMap {{ config.RELEASE_LABEL }}+Score, Chronos)',
    'label': 'Chronos_Combined',
    'dataset_id' : PREPROCESS_TAIGA_ID(virtual_dataset_id, "CRISPRGeneEffect"),
}

# TAIGA_PREPROCESSOR_INCLUDE "xrefs-public.template"