testing_rule_chain.json

{
  "body":
  [
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['AF']",
        "value": "{'ubkg_code': 'C200001', 'assaytype': 'AF', 'dir-schema': 'af-v0', 'tbl-schema': 'af-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'Auto-fluorescence Microscopy', 'dataset-type': 'Auto-fluorescence' }",
        "rule_description": "non-DCWG primary AF"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['AF_pyramid']",
        "value": "{'ubkg_code': 'C200010', 'assaytype': 'AF_pyramid', 'vitessce-hints': ['pyramid', 'is_support'], 'primary': false, 'contains-pii': false, 'description': 'Auto-fluorescence Microscopy [Image Pyramid]', 'pipeline-shorthand': 'Image Pyramid'}",
        "rule_description": "derived AF_pyramid"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['ATACseq-bulk', 'bulkATACseq']",
        "value": "{'ubkg_code': 'C200020', 'assaytype': 'ATACseq-bulk', 'dir-schema': 'bulkatacseq-v0', 'tbl-schema': 'bulkatacseq-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': true, 'primary': true, 'description': 'Bulk ATACseq', 'dataset-type': 'ATACseq' }",
        "rule_description": "non-DCWG primary ATACseq-bulk"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['bulk_atacseq']",
        "value": "{'ubkg_code': 'C200030', 'assaytype': 'bulk_atacseq', 'vitessce-hints': [], 'primary': false, 'contains-pii': false, 'description': 'Bulk ATACseq [BWA + MACS2]', 'pipeline-shorthand': 'BWA + MACS2'}",
        "rule_description": "derived bulk_atacseq"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['cell-dive', 'cell DIVE', 'Cell DIVE']",
        "value": "{'ubkg_code': 'C200040', 'assaytype': 'cell-dive', 'dir-schema': 'celldive-v0', 'tbl-schema': 'celldive-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'Cell DIVE', 'dataset-type': 'Cell DIVE' }",
        "rule_description": "non-DCWG primary cell-dive"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['celldive_deepcell']",
        "value": "{'ubkg_code': 'C200050', 'assaytype': 'celldive_deepcell', 'vitessce-hints': ['sprm', 'anndata', 'is_image', 'is_tiled'], 'primary': false, 'contains-pii': false, 'description': 'Cell DIVE [DeepCell + SPRM]', 'pipeline-shorthand': 'DeepCell + SPRM'}",
        "rule_description": "derived celldive_deepcell"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['CODEX']",
        "value": "{'ubkg_code': 'C200060', 'assaytype': 'CODEX', 'dir-schema': 'codex-v1', 'tbl-schema': 'codex-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'CODEX', 'dataset-type': 'CODEX' }",
        "rule_description": "non-DCWG primary CODEX"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['CODEX2']",
        "value": "{'ubkg_code': 'C200070', 'assaytype': 'CODEX2', 'dir-schema': 'codex-v1', 'tbl-schema': 'codex-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'CODEX (CODEX2 assay type)', 'dataset-type': 'CODEX' }",
        "rule_description": "non-DCWG primary CODEX2"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['codex_cytokit_v1'] and [elt for elt in dag_provenance_list if elt =~~ 'anndata']",
        "value": "{'ubkg_code': 'C200080', 'assaytype': 'codex_cytokit_v1', 'vitessce-hints': ['codex', 'is_image', 'is_tiled', 'anndata'], 'primary': false, 'contains-pii': false, 'description': 'CODEX [Cytokit + SPRM]', 'pipeline-shorthand': 'Cytokit + SPRM'}",
        "rule_description": "derived codex_cytokit_v1"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['codex_cytokit_v1'] and not [elt for elt in dag_provenance_list if elt =~~ 'anndata']",
        "value": "{'ubkg_code': 'C200090', 'assaytype': 'codex_cytokit_v1', 'vitessce-hints': ['codex', 'is_image', 'is_tiled', 'json_based'], 'primary': false, 'contains-pii': false, 'description': 'CODEX [Cytokit + SPRM]', 'pipeline-shorthand': 'Cytokit + SPRM'}",
        "rule_description": "derived codex_cytokit_v1 json-based"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['codex_cytokit']",
        "value": "{'ubkg_code': 'C200100', 'assaytype': 'codex_cytokit', 'vitessce-hints': ['sprm', 'anndata', 'is_image', 'is_tiled'], 'primary': false, 'contains-pii': false, 'description': 'CODEX [Cytokit + SPRM]', 'pipeline-shorthand': 'Cytokit + SPRM'}",
        "rule_description": "derived codex_cytokit"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['DART-FISH']",
        "value": "{'ubkg_code': 'C200110', 'assaytype': 'DART-FISH',   'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'DART-FISH', 'dataset-type': 'DARTFish' }",
        "rule_description": "non-DCWG primary DART-FISH"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['DESI', 'DESI-IMS', 'DESI IMS']",
        "value": "{'ubkg_code': 'C200120', 'assaytype': 'DESI', 'dir-schema': 'ims-v0', 'tbl-schema': 'ims-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'DESI', 'dataset-type': 'DESI' }",
        "rule_description": "non-DCWG primary DESI"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['DESI_pyramid']",
        "value": "{'ubkg_code': 'C200130', 'assaytype': 'DESI_pyramid', 'vitessce-hints': ['pyramid', 'is_support'], 'primary': false, 'contains-pii': false, 'description': 'DESI [Image Pyramid]', 'pipeline-shorthand': 'Image Pyramid'}",
        "rule_description": "derived DESI_pyramid"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['image_pyramid', 'Image Pyramid', 'ometiff_pyramid_ims', 'ometiff_pyramid']",
        "value": "{'ubkg_code': 'C200140', 'assaytype': 'image_pyramid', 'vitessce-hints': ['is_image', 'pyramid', 'is_support'], 'primary': false, 'contains-pii': false, 'description': 'Image Pyramid', 'pipeline-shorthand': ''}",
        "rule_description": "derived image_pyramid"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['IMC2D', 'IMC', '2D-IMC', 'Imaging Mass Cytometry']",
        "value": "{'ubkg_code': 'C200150', 'assaytype': 'IMC2D', 'dir-schema': 'imc-v0', 'tbl-schema': 'imc-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': '2D Imaging Mass Cytometry', 'dataset-type': '2D Imaging Mass Cytometry' }",
        "rule_description": "non-DCWG primary IMC2D"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['IMC3D', '3D-IMC', '3D Imaging Mass Cytometry']",
        "value": "{'ubkg_code': 'C200160', 'assaytype': 'IMC3D', 'dir-schema': 'imc3d-v0', 'tbl-schema': 'imc3d-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': '3D Imaging Mass Cytometry', 'dataset-type': '3D Imaging Mass Cytometry' }",
        "rule_description": "non-DCWG primary IMC3D"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['IMC2D_pyramid', 'IMC_pyramid']",
        "value": "{'ubkg_code': 'C200170', 'assaytype': 'IMC2D_pyramid', 'vitessce-hints': ['pyramid', 'is_support'], 'primary': false, 'contains-pii': false, 'description': '2D Imaging Mass Cytometry [Image Pyramid]', 'pipeline-shorthand': 'Image Pyramid'}",
        "rule_description": "derived IMC2D_pyramid"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['IMC3D_pyramid', '3D-IMC_pyramid']",
        "value": "{'ubkg_code': 'C200180', 'assaytype': 'IMC3D_pyramid', 'vitessce-hints': ['pyramid', 'is_support'], 'primary': false, 'contains-pii': false, 'description': '3D Imaging Mass Cytometry [Image Pyramid]', 'pipeline-shorthand': 'Image Pyramid'}",
        "rule_description": "derived IMC3D_pyramid"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['lc-ms_label-free']",
        "value": "{'ubkg_code': 'C200190', 'assaytype': 'lc-ms_label-free',   'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'Label-free LC-MS', 'dataset-type': 'LC-MS' }",
        "rule_description": "non-DCWG primary lc-ms_label-free"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['lc-ms_labeled']",
        "value": "{'ubkg_code': 'C200200', 'assaytype': 'lc-ms_labeled',   'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'Labeled LC-MS', 'dataset-type': 'LC-MS' }",
        "rule_description": "non-DCWG primary lc-ms_labeled"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['lc-ms-ms_label-free', 'LC-MS/MS (label-free proteomics)']",
        "value": "{'ubkg_code': 'C200210', 'assaytype': 'lc-ms-ms_label-free',   'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'Label-free LC-MS/MS', 'dataset-type': 'LC-MS' }",
        "rule_description": "non-DCWG primary lc-ms-ms_label-free"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['lc-ms-ms_labeled']",
        "value": "{'ubkg_code': 'C200220', 'assaytype': 'lc-ms-ms_labeled',   'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'Labeled LC-MS/MS', 'dataset-type': 'LC-MS' }",
        "rule_description": "non-DCWG primary lc-ms-ms_labeled"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['LC-MS-untargeted']",
        "value": "{'ubkg_code': 'C200230', 'assaytype': 'LC-MS-untargeted',   'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'Untargeted LC-MS', 'dataset-type': 'LC-MS' }",
        "rule_description": "non-DCWG primary LC-MS-untargeted"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['Lightsheet', 'Light Sheet']",
        "value": "{'ubkg_code': 'C200240', 'assaytype': 'Lightsheet', 'dir-schema': 'lightsheet-v1', 'tbl-schema': 'lightsheet-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'Light Sheet Microscopy', 'dataset-type': 'Light Sheet' }",
        "rule_description": "non-DCWG primary Lightsheet"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['MALDI-IMS', 'MALDI-IMS-neg', 'MALDI-IMS-pos', 'IMS negative', 'IMS positive']",
        "value": "{'ubkg_code': 'C200250', 'assaytype': 'MALDI-IMS', 'dir-schema': 'ims-v0', 'tbl-schema': 'ims-v'+version.to_str, 'vitessce-hints': ['maldi'], 'contains-pii': false, 'primary': true, 'description': 'MALDI IMS', 'dataset-type': 'MALDI' }",
        "rule_description": "non-DCWG primary MALDI-IMS"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['MALDI-IMS_pyramid', 'MALDI-IMS-neg_pyramid', 'MALDI-IMS-pos_pyramid']",
        "value": "{'ubkg_code': 'C200260', 'assaytype': 'MALDI-IMS_pyramid', 'vitessce-hints': ['pyramid', 'maldi', 'is_support'], 'primary': false, 'contains-pii': false, 'description': 'MALDI IMS [Image Pyramid]', 'pipeline-shorthand': 'Image Pyramid'}",
        "rule_description": "derived MALDI-IMS_pyramid"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['MIBI', 'Multiplex Ion Beam Imaging', 'mibi']",
        "value": "{'ubkg_code': 'C200270', 'assaytype': 'MIBI', 'dir-schema': 'mibi-v0', 'tbl-schema': 'mibi-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'Multiplex Ion Beam Imaging', 'dataset-type': 'MIBI' }",
        "rule_description": "non-DCWG primary MIBI"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['mibi_deepcell']",
        "value": "{'ubkg_code': 'C200280', 'assaytype': 'mibi_deepcell', 'vitessce-hints': ['is_image', 'is_tiled', 'sprm', 'anndata'], 'primary': false, 'contains-pii': false, 'description': 'Multiplex Ion Beam Imaging [DeepCell + SPRM]', 'pipeline-shorthand': 'DeepCell + SPRM'}",
        "rule_description": "derived mibi_deepcell"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['NanoDESI']",
        "value": "{'ubkg_code': 'C200290', 'assaytype': 'NanoDESI', 'dir-schema': 'nano-v0', 'tbl-schema': 'nano-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'NanoDESI', 'dataset-type': 'DESI' }",
        "rule_description": "non-DCWG primary NanoDESI"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['NanoDESI_pyramid']",
        "value": "{'ubkg_code': 'C200300', 'assaytype': 'NanoDESI_pyramid', 'vitessce-hints': ['pyramid', 'is_support'], 'primary': false, 'contains-pii': false, 'description': 'NanoDESI [Image Pyramid]', 'pipeline-shorthand': 'Image Pyramid'}",
        "rule_description": "derived NanoDESI_pyramid"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['NanoPOTS']",
        "value": "{'ubkg_code': 'C200310', 'assaytype': 'NanoPOTS', 'dir-schema': 'nano-v0', 'tbl-schema': 'nano-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'NanoPOTS', 'dataset-type': 'UNKNOWN' }",
        "rule_description": "non-DCWG primary NanoPOTS"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['NanoPOTS_pyramid']",
        "value": "{'ubkg_code': 'C200320', 'assaytype': 'NanoPOTS_pyramid', 'vitessce-hints': ['pyramid', 'is_support'], 'primary': false, 'contains-pii': false, 'description': 'NanoPOTS [Image Pyramid]', 'pipeline-shorthand': 'Image Pyramid'}",
        "rule_description": "derived NanoPOTS_pyramid"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['MxIF']",
        "value": "{'ubkg_code': 'C200330', 'assaytype': 'MxIF', 'dir-schema': 'mxif-v0', 'tbl-schema': 'mxif-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'Multiplexed IF Microscopy', 'dataset-type': 'UNKNOWN' }",
        "rule_description": "non-DCWG primary MxIF"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['MxIF_pyramid']",
        "value": "{'ubkg_code': 'C200340', 'assaytype': 'MxIF_pyramid', 'vitessce-hints': ['pyramid', 'is_support'], 'primary': false, 'contains-pii': false, 'description': 'Multiplexed IF Microscopy [Image Pyramid]', 'pipeline-shorthand': 'Image Pyramid'}",
        "rule_description": "derived MxIF_pyramid"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['PAS', 'PAS microscopy']",
        "value": "{'ubkg_code': 'C200350', 'assaytype': 'PAS', 'dir-schema': 'stained-v0', 'tbl-schema': 'stained-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'PAS Stained Microscopy', 'dataset-type': 'Histology' }",
        "rule_description": "non-DCWG primary PAS"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['PAS_pyramid']",
        "value": "{'ubkg_code': 'C200360', 'assaytype': 'PAS_pyramid', 'vitessce-hints': ['pyramid', 'is_support'], 'primary': false, 'contains-pii': false, 'description': 'PAS Stained Microscopy [Image Pyramid]', 'pipeline-shorthand': 'Image Pyramid'}",
        "rule_description": "derived PAS_pyramid"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['pas_ftu_segmentation']",
        "value": "{'ubkg_code': 'C200370', 'assaytype': 'pas_ftu_segmentation', 'vitessce-hints': [], 'primary': false, 'contains-pii': false, 'description': 'PAS Stained Microscopy [Kaggle-1 Glomerulus Segmentation]', 'pipeline-shorthand': 'Kaggle-1 Glomerulus Segmentation'}",
        "rule_description": "derived pas_ftu_segmentation"
    },
    {
        "type": "match",
        "match": "is_publication",
        "value": "{'ubkg_code': 'C200380', 'assaytype': 'publication', 'vitessce-hints': [], 'primary': false, 'contains-pii': false, 'description': 'Publication Data'}",
        "rule_description": "derived publication"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['publication_ancillary', 'gen_pub_ancillary']",
        "value": "{'ubkg_code': 'C200390', 'assaytype': 'publication_ancillary', 'vitessce-hints': ['json', 'is_support'], 'primary': false, 'contains-pii': false, 'description': 'Publication Data [ancillary]', 'pipeline-shorthand': 'ancillary'}",
        "rule_description": "derived publication_ancillary"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['bulk-RNA', 'bulk RNA']",
        "value": "{'ubkg_code': 'C200400', 'assaytype': 'bulk-RNA', 'dir-schema': 'bulkrnaseq-v0', 'tbl-schema': 'bulkrnaseq-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': true, 'primary': true, 'description': 'Bulk RNAseq', 'dataset-type': 'RNAseq' }",
        "rule_description": "non-DCWG primary bulk-RNA"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['salmon_rnaseq_bulk']",
        "value": "{'ubkg_code': 'C200410', 'assaytype': 'salmon_rnaseq_bulk', 'vitessce-hints': [], 'primary': false, 'contains-pii': false, 'description': 'Bulk RNAseq [Salmon]', 'pipeline-shorthand': 'Salmon'}",
        "rule_description": "derived salmon_rnaseq_bulk"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['SNARE-ATACseq2', 'SNAREseq', 'SNARE-seq2', 'SNARE2-ATACseq']",
        "value": "{'ubkg_code': 'C200420', 'assaytype': 'SNARE-ATACseq2', 'dir-schema': 'scatacseq-v0', 'tbl-schema': 'scatacseq-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': true, 'primary': true, 'description': 'snATACseq (SNAREseq2)', 'dataset-type': 'ATACseq' }",
        "rule_description": "non-DCWG primary SNARE-ATACseq2"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['SNARE-RNAseq2', 'SNARE2-RNAseq']",
        "value": "{'ubkg_code': 'C200430', 'assaytype': 'SNARE-RNAseq2', 'dir-schema': 'scrnaseq-v0', 'tbl-schema': 'scrnaseq-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': true, 'primary': true, 'description': 'snRNAseq (SNAREseq2)', 'dataset-type': 'RNAseq' }",
        "rule_description": "non-DCWG primary SNARE-RNAseq2"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_lab_processed and data_types[0] in ['sc_atac_seq_snare_lab']",
        "value": "{'ubkg_code': 'C200440', 'assaytype': 'sc_atac_seq_snare_lab', 'vitessce-hints': [], 'primary': false, 'contains-pii': false, 'description': 'snATACseq (SNAREseq2) [Lab Processed]'}",
        "rule_description": "derived sc_atac_seq_snare_lab"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_lab_processed and data_types[0] in ['sc_rna_seq_snare_lab']",
        "value": "{'ubkg_code': 'C200450', 'assaytype': 'sc_rna_seq_snare_lab', 'vitessce-hints': [], 'primary': false, 'contains-pii': false, 'description': 'snRNAseq (SNAREseq2) [Lab Processed]'}",
        "rule_description": "derived sc_rna_seq_snare_lab"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['salmon_rnaseq_snareseq']",
        "value": "{'ubkg_code': 'C200460', 'assaytype': 'salmon_rnaseq_snareseq', 'vitessce-hints': ['is_sc', 'rna'], 'primary': false, 'contains-pii': false, 'description': 'snRNAseq (SNAREseq2) [Salmon]', 'pipeline-shorthand': 'Salmon'}",
        "rule_description": "derived salmon_rnaseq_snareseq"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['sc_atac_seq_snare']",
        "value": "{'ubkg_code': 'C200470', 'assaytype': 'sc_atac_seq_snare', 'vitessce-hints': ['is_sc', 'atac'], 'primary': false, 'contains-pii': false, 'description': 'snATACseq (SNAREseq2) [SnapATAC]', 'pipeline-shorthand': 'SnapATAC'}",
        "rule_description": "derived sc_atac_seq_snare"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['scRNAseq-10xGenomics-v2']",
        "value": "{'ubkg_code': 'C200480', 'assaytype': 'scRNAseq-10xGenomics-v2', 'dir-schema': 'scrnaseq-v0', 'tbl-schema': 'scrnaseq-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': true, 'primary': true, 'description': 'scRNAseq (10x Genomics v2)', 'dataset-type': 'RNAseq' }",
        "rule_description": "non-DCWG primary scRNAseq-10xGenomics-v2"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['scRNAseq-10xGenomics-v3', 'scRNA-Seq(10xGenomics)', 'scRNA-Seq-10x', 'scRNAseq-10xGenomics']",
        "value": "{'ubkg_code': 'C200490', 'assaytype': 'scRNAseq-10xGenomics-v3', 'dir-schema': 'scrnaseq-v0', 'tbl-schema': 'scrnaseq-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': true, 'primary': true, 'description': 'scRNAseq (10x Genomics v3)', 'dataset-type': 'RNAseq' }",
        "rule_description": "non-DCWG primary scRNAseq-10xGenomics-v3"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['salmon_rnaseq_10x', 'salmon_rnaseq_10x_v2'] and [elt for elt in dag_provenance_list if elt =~~ 'anndata']",
        "value": "{'ubkg_code': 'C200500', 'assaytype': 'salmon_rnaseq_10x', 'vitessce-hints': ['is_sc', 'rna'], 'primary': false, 'contains-pii': false, 'description': 'scRNAseq (10x Genomics) [Salmon]', 'pipeline-shorthand': 'Salmon'}",
        "rule_description": "derived salmon_rnaseq_10x"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['salmon_rnaseq_10x', 'salmon_rnaseq_10x_v2'] and not [elt for elt in dag_provenance_list if elt =~~ 'anndata']",
        "value": "{'ubkg_code': 'C202010', 'assaytype': 'salmon_rnaseq_10x', 'vitessce-hints': ['is_sc', 'rna', 'json_based'], 'primary': false, 'contains-pii': false, 'description': 'scRNAseq (10x Genomics) [Salmon]', 'pipeline-shorthand': 'Salmon'}",
        "rule_description": "derived salmon_rnaseq_10x json-based"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['sciATACseq']",
        "value": "{'ubkg_code': 'C200510', 'assaytype': 'sciATACseq', 'dir-schema': 'scatacseq-v0', 'tbl-schema': 'scatacseq-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': true, 'primary': true, 'description': 'sciATACseq', 'dataset-type': 'ATACseq' }",
        "rule_description": "non-DCWG primary sciATACseq"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['sc_atac_seq_sci']",
        "value": "{'ubkg_code': 'C200520', 'assaytype': 'sc_atac_seq_sci', 'vitessce-hints': ['is_sc', 'atac'], 'primary': false, 'contains-pii': false, 'description': 'sciATACseq [SnapATAC]', 'pipeline-shorthand': 'SnapATAC'}",
        "rule_description": "derived sc_atac_seq_sci"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['sciRNAseq']",
        "value": "{'ubkg_code': 'C200530', 'assaytype': 'sciRNAseq', 'dir-schema': 'scrnaseq-v0', 'tbl-schema': 'scrnaseq-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': true, 'primary': true, 'description': 'sciRNAseq', 'dataset-type': 'RNAseq' }",
        "rule_description": "non-DCWG primary sciRNAseq"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['salmon_rnaseq_sciseq']",
        "value": "{'ubkg_code': 'C200540', 'assaytype': 'salmon_rnaseq_sciseq', 'vitessce-hints': ['is_sc', 'rna'], 'primary': false, 'contains-pii': false, 'description': 'sciRNAseq [Salmon]', 'pipeline-shorthand': 'Salmon'}",
        "rule_description": "derived salmon_rnaseq_sciseq"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['seqFish', 'seqFISH']",
        "value": "{'ubkg_code': 'C200550', 'assaytype': 'seqFish', 'dir-schema': 'seqfish-v0', 'tbl-schema': 'seqfish-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'seqFISH', 'dataset-type': 'seqFISH' }",
        "rule_description": "non-DCWG primary seqFish"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['seqFish_pyramid']",
        "value": "{'ubkg_code': 'C200560', 'assaytype': 'seqFish_pyramid', 'vitessce-hints': ['pyramid', 'is_support'], 'primary': false, 'contains-pii': false, 'description': 'seqFISH [Image Pyramid]', 'pipeline-shorthand': 'Image Pyramid'}",
        "rule_description": "derived seqFish_pyramid"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_lab_processed and data_types[0] in ['seqFish_lab_processed']",
        "value": "{'ubkg_code': 'C200570', 'assaytype': 'seqFish_lab_processed', 'vitessce-hints': [], 'primary': false, 'contains-pii': false, 'description': 'seqFISH [Lab Processed]'}",
        "rule_description": "derived seqFish_lab_processed"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['SIMS-IMS', 'SIMS']",
        "value": "{'ubkg_code': 'C200580', 'assaytype': 'SIMS-IMS', 'dir-schema': 'ims-v0', 'tbl-schema': 'ims-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'SIMS-IMS', 'dataset-type': 'SIMS' }",
        "rule_description": "non-DCWG primary SIMS-IMS"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['snATACseq']",
        "value": "{'ubkg_code': 'C200590', 'assaytype': 'snATACseq', 'dir-schema': 'scatacseq-v0', 'tbl-schema': 'scatacseq-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': true, 'primary': true, 'description': 'snATACseq', 'dataset-type': 'ATACseq' }",
        "rule_description": "non-DCWG primary snATACseq"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['sn_atac_seq', 'sn_atac_seq_multiome_10x', 'sc_atac_seq_sn', 'sc_atac_seq_multiome_10x']",
        "value": "{'ubkg_code': 'C200600', 'assaytype': 'sn_atac_seq', 'vitessce-hints': ['is_sc', 'atac'], 'primary': false, 'contains-pii': false, 'description': 'snATACseq [SnapATAC]', 'pipeline-shorthand': 'SnapATAC'}",
        "rule_description": "derived sn_atac_seq"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['snRNAseq-10xGenomics-v2', 'snRNAseq-v2']",
        "value": "{'ubkg_code': 'C200610', 'assaytype': 'snRNAseq-10xGenomics-v2', 'dir-schema': 'scrnaseq-v0', 'tbl-schema': 'scrnaseq-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': true, 'primary': true, 'description': 'snRNAseq (10x Genomics v2)', 'dataset-type': 'RNAseq' }",
        "rule_description": "non-DCWG primary snRNAseq-10xGenomics-v2"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['snRNAseq-10xGenomics-v3', 'snRNAseq', 'snRNAseq-v3']",
        "value": "{'ubkg_code': 'C200620', 'assaytype': 'snRNAseq-10xGenomics-v3', 'dir-schema': 'scrnaseq-v0', 'tbl-schema': 'scrnaseq-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': true, 'primary': true, 'description': 'snRNAseq (10x Genomics v3)', 'dataset-type': 'RNAseq' }",
        "rule_description": "non-DCWG primary snRNAseq-10xGenomics-v3"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['salmon_sn_rnaseq_10x', 'salmon_rnaseq_10x_sn', 'salmon_rnaseq_10x_v2_sn'] and [elt for elt in dag_provenance_list if elt =~~ 'anndata']",
        "value": "{'ubkg_code': 'C200630', 'assaytype': 'salmon_sn_rnaseq_10x', 'vitessce-hints': ['is_sc', 'rna'], 'primary': false, 'contains-pii': false, 'description': 'snRNAseq (10x Genomics) [Salmon]', 'pipeline-shorthand': 'Salmon'}",
        "rule_description": "derived salmon_sn_rnaseq_10x"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['salmon_sn_rnaseq_10x', 'salmon_rnaseq_10x_sn', 'salmon_rnaseq_10x_v2_sn'] and not [elt for elt in dag_provenance_list if elt =~~ 'anndata']",
        "value": "{'ubkg_code': 'C202000', 'assaytype': 'salmon_sn_rnaseq_10x', 'vitessce-hints': ['is_sc', 'rna', 'json_based'], 'primary': false, 'contains-pii': false, 'description': 'snRNAseq (10x Genomics) [Salmon]', 'pipeline-shorthand': 'Salmon'}",
        "rule_description": "derived salmon_sn_rnaseq_10x json-based"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['Slide-seq']",
        "value": "{'ubkg_code': 'C200640', 'assaytype': 'Slide-seq', 'dir-schema': 'slideseq-v0', 'tbl-schema': 'slideseq-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': true, 'primary': true, 'description': 'Slideseq', 'dataset-type': 'Slideseq' }",
        "rule_description": "non-DCWG primary Slide-seq"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['salmon_rnaseq_slideseq']",
        "value": "{'ubkg_code': 'C200650', 'assaytype': 'salmon_rnaseq_slideseq', 'vitessce-hints': ['is_sc', 'rna'], 'primary': false, 'contains-pii': false, 'description': 'Slideseq [Salmon]', 'pipeline-shorthand': 'Salmon'}",
        "rule_description": "derived salmon_rnaseq_slideseq"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['Targeted-Shotgun-LC-MS']",
        "value": "{'ubkg_code': 'C200660', 'assaytype': 'Targeted-Shotgun-LC-MS',   'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'Targeted Shotgun / Flow-injection LC-MS', 'dataset-type': 'LC-MS' }",
        "rule_description": "non-DCWG primary Targeted-Shotgun-LC-MS"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['TMT-LC-MS', 'TMT (proteomics)']",
        "value": "{'ubkg_code': 'C200670', 'assaytype': 'TMT-LC-MS',   'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'TMT LC-MS', 'dataset-type': 'LC-MS' }",
        "rule_description": "non-DCWG primary TMT-LC-MS"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['WGS', 'Whole Transcriptome Sequencing', 'Whole Genome Sequencing']",
        "value": "{'ubkg_code': 'C200680', 'assaytype': 'WGS', 'dir-schema': 'wgs-v0', 'tbl-schema': 'wgs-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': true, 'primary': true, 'description': 'Whole Genome Sequencing', 'dataset-type': 'WGS' }",
        "rule_description": "non-DCWG primary WGS"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['LC-MS', 'LC-MS (metabolomics)']",
        "value": "{'ubkg_code': 'C200690', 'assaytype': 'LC-MS', 'dir-schema': 'lcms-v0', 'tbl-schema': 'lcms-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'LC-MS', 'dataset-type': 'LC-MS' }",
        "rule_description": "non-DCWG primary LC-MS"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['MS', 'MS (shotgun lipidomics)']",
        "value": "{'ubkg_code': 'C201190', 'assaytype': 'MS', 'dir-schema': 'lcms-v0', 'tbl-schema': 'lcms-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'MS', 'dataset-type': 'LC-MS' }",
        "rule_description": "non-DCWG primary MS"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['LC-MS_bottom_up', 'LC-MS Bottom-Up']",
        "value": "{'ubkg_code': 'C200700', 'assaytype': 'LC-MS_bottom_up', 'dir-schema': 'lcms-v0', 'tbl-schema': 'lcms-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'LC-MS Bottom Up', 'dataset-type': 'LC-MS' }",
        "rule_description": "non-DCWG primary LC-MS_bottom_up"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['MS_bottom_up', 'MS Bottom-Up']",
        "value": "{'ubkg_code': 'C200710', 'assaytype': 'MS_bottom_up', 'dir-schema': 'lcms-v0', 'tbl-schema': 'lcms-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'MS Bottom Up', 'dataset-type': 'LC-MS' }",
        "rule_description": "non-DCWG primary MS_bottom_up"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['LC-MS_top_down', 'LC-MS Top-Down']",
        "value": "{'ubkg_code': 'C200720', 'assaytype': 'LC-MS_top_down', 'dir-schema': 'lcms-v0', 'tbl-schema': 'lcms-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'LC-MS Top Down', 'dataset-type': 'LC-MS' }",
        "rule_description": "non-DCWG primary LC-MS_top_down"
    },
    {
        "type": "match",
        "match": "not_dcwg and is_primary and assay_type in ['MS_top_down', 'MS Top-Down']",
        "value": "{'ubkg_code': 'C200730', 'assaytype': 'MS_top_down', 'dir-schema': 'lcms-v0', 'tbl-schema': 'lcms-v'+version.to_str, 'vitessce-hints': [], 'contains-pii': false, 'primary': true, 'description': 'MS Top Down', 'dataset-type': 'LC-MS' }",
        "rule_description": "non-DCWG primary MS_top_down"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Visium (no probes)'",
        "value": "{'ubkg_code': 'C200740', 'assaytype': 'visium-no-probes', 'vitessce-hints': [], 'dir-schema': 'visium-no-probes-v3', 'description': 'Visium (no probes)', 'contains-pii': true, 'primary': true, 'dataset-type': 'Visium (no probes)', 'must-contain': ['Histology','RNAseq'], 'is-multi-assay': true}",
        "rule_description": "DCWG visium-no-probes"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Visium (with probes)'",
        "value": "{'ubkg_code': 'C200750', 'assaytype': 'visium-with-probes', 'vitessce-hints': [], 'dir-schema': 'visium-with-probes-v3', 'description': 'Visium (with probes)', 'contains-pii': false, 'primary': true, 'dataset-type': 'Visium (with probes)', 'must-contain': ['Histology','RNAseq (with probes)'], 'is-multi-assay': true}",
        "rule_description": "DCWG visium-with-probes"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'GeoMx (NGS)'",
        "value": "{'ubkg_code': 'C200760', 'assaytype': 'geomx_ngs?', 'vitessce-hints': [], 'dir-schema': 'geomx-ngs-v2', 'description': 'GeoMx (NGS)', 'contains-pii': true, 'primary': true, 'dataset-type': 'GeoMx (NGS)', 'must-contain': ['RNAseq (with probes)'], 'is-multi-assay': true}",
        "rule_description": "DCWG geomx_ngs?"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == '10X Multiome'",
        "value": "{'ubkg_code': 'C200770', 'assaytype': '10x-multiome', 'vitessce-hints': [], 'dir-schema': '10x-multiome-v2', 'description': '10x Multiome', 'contains-pii': true, 'primary': true, 'dataset-type': '10X Multiome', 'must-contain': ['RNAseq','ATACseq'], 'is-multi-assay': true}",
        "rule_description": "DCWG 10x-multiome"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'SNARE-seq2'",
        "value": "{'ubkg_code': 'C200780', 'assaytype': 'multiome-snare-seq2', 'vitessce-hints': [], 'dir-schema': 'snareseq2-v2', 'description': 'SNAREseq2', 'contains-pii': true, 'primary': true, 'dataset-type': 'SNARE-seq2', 'must-contain': ['RNAseq','ATACseq'], 'is-multi-assay': true}",
        "rule_description": "DCWG multiome-snare-seq2"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'RNAseq'  and assay_input_entity == 'single cell' and barcode_read =~~ 'Not applicable' and barcode_size == 40 and barcode_offset == 'Not applicable' and umi_read =~~ 'Not applicable' and umi_size == 8 and umi_offset == 'Not applicable'",
        "value": "{'ubkg_code': 'C200790', 'assaytype': 'sciRNAseq', 'vitessce-hints': [], 'dir-schema': 'rnaseq-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'RNAseq', 'description': 'sciRNAseq'}",
        "rule_description": "DCWG sciRNAseq"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'RNAseq'  and assay_input_entity == 'single nucleus' and barcode_read =~~ 'Read 2' and barcode_size == '8,8,8' and barcode_offset == '10,48,86' and umi_read =~~ 'Read 2' and umi_size == 10 and umi_offset == 0",
        "value": "{'ubkg_code': 'C200800', 'assaytype': 'SNARE-RNAseq2', 'vitessce-hints': [], 'dir-schema': 'rnaseq-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'RNAseq', 'description': 'snRNAseq (SNAREseq2)'}",
        "rule_description": "DCWG SNARE-RNAseq2"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'RNAseq' and assay_input_entity == 'spot' and barcode_read =~~ 'Read 1' and barcode_size == 16 and umi_read =~~ 'Read 1' and umi_size == 12 and ((barcode_offset==0 and umi_offset==16) or (barcode_offset==20 and umi_offset==36))",
        "value": "{'ubkg_code': 'C200810', 'assaytype': 'rnaseq-visium-no-probes', 'vitessce-hints': [], 'dir-schema': 'rnaseq-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'RNAseq', 'description': 'Capture bead RNAseq (10x Genomics v3)'}",
        "rule_description": "DCWG rnaseq-visium-no-probes"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'RNAseq (with probes)' and oligo_probe_panel in ['10x Genomics; Visium Human Transcriptome Probe Kit v2 - Small; PN 1000466', '10x Genomics; Visium Human Transcriptome Probe Kit-Small; PN 1000363', '10x Genomics; Visium Human Transcriptome Probe Kit-Large; PN 1000364', '10x Genomics; Visium Mouse Transcriptome Probe Kit - Small; PN 1000365', '10x Genomics; Visium Mouse Transcriptome Probes v2 - small; PN 1000667'] and assay_input_entity == 'spot' and barcode_read =~~ 'Read 1' and barcode_size == 16 and barcode_offset == 0 and umi_read =~~ 'Read 1' and umi_size == 12 and umi_offset == 16",
        "value": "{'ubkg_code': 'C200820', 'assaytype': 'scRNAseq-visium-with-probes', 'vitessce-hints': [], 'dir-schema': 'rnaseq-with-probes-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'RNAseq (with probes)', 'description': 'Visium RNAseq (with probes)'}",
        "rule_description": "DCWG scRNAseq-visium-with-probes"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'RNAseq'  and assay_input_entity == 'single cell' and barcode_read =~~ 'Read 1' and barcode_size == 16 and barcode_offset == 0 and umi_read =~~ 'Read 1' and umi_size == 10 and umi_offset == 16",
        "value": "{'ubkg_code': 'C200830', 'assaytype': 'scRNAseq-10xGenomics-v2', 'vitessce-hints': [], 'dir-schema': 'rnaseq-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'RNAseq', 'description': 'scRNAseq (10x Genomics v2)'}",
        "rule_description": "DCWG scRNAseq-10xGenomics-v2"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'RNAseq'  and assay_input_entity == 'single nucleus' and barcode_read =~~ 'Read 1' and barcode_size == 16 and barcode_offset == 0 and umi_read =~~ 'Read 1' and umi_size == 10 and umi_offset == 16",
        "value": "{'ubkg_code': 'C200840', 'assaytype': 'snRNAseq-10xGenomics-v2', 'vitessce-hints': [], 'dir-schema': 'rnaseq-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'RNAseq', 'description': 'snRNAseq (10x Genomics v2)'}",
        "rule_description": "DCWG snRNAseq-10xGenomics-v2"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'RNAseq'  and assay_input_entity == 'single cell' and barcode_read =~~ 'Read 1' and barcode_size == 16 and barcode_offset == 0 and umi_read =~~ 'Read 1' and umi_size == 12 and umi_offset == 16",
        "value": "{'ubkg_code': 'C200850', 'assaytype': 'scRNAseq-10xGenomics-v3', 'vitessce-hints': [], 'dir-schema': 'rnaseq-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'RNAseq', 'description': 'scRNAseq (10x Genomics v3)'}",
        "rule_description": "DCWG scRNAseq-10xGenomics-v3"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'RNAseq'  and assay_input_entity == 'single nucleus' and barcode_read =~~ 'Read 1' and barcode_size == 16 and barcode_offset == 0 and umi_read =~~ 'Read 1' and umi_size == 12 and umi_offset == 16",
        "value": "{'ubkg_code': 'C200860', 'assaytype': 'snRNAseq-10xGenomics-v3', 'vitessce-hints': [], 'dir-schema': 'rnaseq-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'RNAseq', 'description': 'snRNAseq (10x Genomics v3)'}",
        "rule_description": "DCWG snRNAseq-10xGenomics-v3"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'ATACseq'  and assay_input_entity == 'single nucleus' and barcode_read =~~ 'Read 2' and barcode_size == 16 and barcode_offset == 0 and umi_read =~~ 'Not applicable' and umi_size == 'Not applicable' and umi_offset == 'Not applicable'",
        "value": "{'ubkg_code': 'C200870', 'assaytype': 'snATACseq', 'vitessce-hints': [], 'dir-schema': 'atacseq-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'ATACseq', 'description': 'snATACseq'}",
        "rule_description": "DCWG snATACseq"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'ATACseq'  and assay_input_entity == 'single nucleus' and barcode_read =~~ 'Read 2' and barcode_size == '8,8,8' and barcode_offset == '0,38,76' and umi_read =~~ 'Not applicable' and umi_size == 'Not applicable' and umi_offset == 'Not applicable'",
        "value": "{'ubkg_code': 'C200880', 'assaytype': 'SNARE-ATACseq2', 'vitessce-hints': [], 'dir-schema': 'atacseq-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'ATACseq', 'description': 'snATACseq (SNAREseq2)'}",
        "rule_description": "DCWG SNARE-ATACseq2"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'ATACseq'  and assay_input_entity == 'single nucleus' and barcode_read =~~ 'Read 2' and barcode_size == 16 and barcode_offset == 8 and umi_read =~~ 'Not applicable' and umi_size == 'Not applicable' and umi_offset == 'Not applicable'",
        "value": "{'ubkg_code': 'C200890', 'assaytype': 'sn_atac_seq?', 'vitessce-hints': [], 'dir-schema': 'atacseq-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'ATACseq', 'description': 'snATACseq-multiome'}",
        "rule_description": "DCWG sn_atac_seq?"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'RNAseq' and assay_input_entity == 'tissue (bulk)'",
        "value": "{'ubkg_code': 'C200900', 'assaytype': 'bulk-RNA', 'vitessce-hints': [], 'dir-schema': 'rnaseq-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'RNAseq', 'description': 'Bulk RNAseq'}",
        "rule_description": "DCWG bulk-RNA"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'ATACseq' and assay_input_entity == 'tissue (bulk)'",
        "value": "{'ubkg_code': 'C200910', 'assaytype': 'ATACseq-bulk', 'vitessce-hints': [], 'dir-schema': 'atacseq-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'ATACseq', 'description': 'Bulk ATACseq'}",
        "rule_description": "DCWG ATACseq-bulk"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'ATACseq' and assay_input_entity == 'single cell' and barcode_read =~~ 'Not applicable'",
        "value": "{'ubkg_code': 'C200920', 'assaytype': 'sciATACseq', 'vitessce-hints': [], 'dir-schema': 'atacseq-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'ATACseq', 'description': 'sciATACseq'}",
        "rule_description": "DCWG sciATACseq"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Histology'  and stain_name == 'PAS'",
        "value": "{'ubkg_code': 'C200930', 'assaytype': 'PAS', 'vitessce-hints': [], 'dir-schema': 'histology-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'Histology', 'description': 'PAS Stained Microscopy'}",
        "rule_description": "DCWG PAS"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Histology'  and stain_name == 'H&E'",
        "value": "{'ubkg_code': 'C200940', 'assaytype': 'h-and-e', 'vitessce-hints': [], 'dir-schema': 'histology-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'Histology', 'description': 'H&E Stained Microscopy'}",
        "rule_description": "DCWG h-and-e"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'CODEX'",
        "value": "{'ubkg_code': 'C200950', 'assaytype': 'CODEX', 'vitessce-hints': [], 'dir-schema': 'codex-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'CODEX', 'description': 'CODEX'}",
        "rule_description": "DCWG CODEX"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'PhenoCycler'",
        "value": "{'ubkg_code': 'C200960', 'assaytype': 'phenocycler', 'vitessce-hints': [], 'dir-schema': 'phenocycler-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'PhenoCycler', 'description': 'PhenoCycler'}",
        "rule_description": "DCWG phenocycler"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'CycIF'",
        "value": "{'ubkg_code': 'C200970', 'assaytype': 'cycif', 'vitessce-hints': [], 'dir-schema': 'cycif-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'CycIF', 'description': 'CycIF'}",
        "rule_description": "DCWG cycif"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'MERFISH'",
        "value": "{'ubkg_code': 'C200980', 'assaytype': 'merfish', 'vitessce-hints': [], 'dir-schema': 'merfish-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'MERFISH', 'description': 'MERFISH'}",
        "rule_description": "DCWG merfish"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Cell DIVE'",
        "value": "{'ubkg_code': 'C200990', 'assaytype': 'cell-dive', 'vitessce-hints': [], 'dir-schema': 'celldive-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'Cell DIVE', 'description': 'Cell DIVE'}",
        "rule_description": "DCWG cell-dive"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'MALDI'",
        "value": "{'ubkg_code': 'C201000', 'assaytype': 'MALDI-IMS', 'vitessce-hints': [], 'dir-schema': 'maldi-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'MALDI', 'description': 'MALDI IMS'}",
        "rule_description": "DCWG MALDI-IMS"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'SIMS'",
        "value": "{'ubkg_code': 'C201010', 'assaytype': 'SIMS-IMS', 'vitessce-hints': [], 'dir-schema': 'sims-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'SIMS', 'description': 'SIMS-IMS'}",
        "rule_description": "DCWG SIMS-IMS"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'DESI'",
        "value": "{'ubkg_code': 'C201020', 'assaytype': 'DESI-IMS', 'vitessce-hints': [], 'dir-schema': 'desi-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'DESI', 'description': 'DESI'}",
        "rule_description": "DCWG DESI-IMS"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'MIBI'",
        "value": "{'ubkg_code': 'C201030', 'assaytype': 'MIBI', 'vitessce-hints': [], 'dir-schema': 'mibi-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'MIBI', 'description': 'Multiplex Ion Beam Imaging'}",
        "rule_description": "DCWG MIBI"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == '2D Imaging Mass Cytometry'",
        "value": "{'ubkg_code': 'C201040', 'assaytype': 'IMC2D', 'vitessce-hints': [], 'dir-schema': 'imc-2d-v2', 'contains-pii': false, 'primary': true, 'dataset-type': '2D Imaging Mass Cytometry', 'description': '2D Imaging Mass Cytometry'}",
        "rule_description": "DCWG IMC2D"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'LC-MS'",
        "value": "{'ubkg_code': 'C201050', 'assaytype': 'LC-MS', 'vitessce-hints': [], 'dir-schema': 'lcms-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'LC-MS', 'description': 'LC-MS'}",
        "rule_description": "DCWG LC-MS"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'nanoSPLITS'",
        "value": "{'ubkg_code': 'C201060', 'assaytype': 'nano-splits', 'vitessce-hints': [], 'dir-schema': 'nano-splits-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'nanoSPLITS', 'description': 'nanoSPLITS'}",
        "rule_description": "DCWG nano-splits"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Auto-fluorescence'",
        "value": "{'ubkg_code': 'C201070', 'assaytype': 'AF', 'vitessce-hints': [], 'dir-schema': 'af-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'Auto-fluorescence', 'description': 'Auto-fluorescence Microscopy'}",
        "rule_description": "DCWG AF"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Light Sheet'",
        "value": "{'ubkg_code': 'C201080', 'assaytype': 'Lightsheet', 'vitessce-hints': [], 'dir-schema': 'lightsheet-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'Light Sheet', 'description': 'Light Sheet Microscopy'}",
        "rule_description": "DCWG Lightsheet"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Confocal'",
        "value": "{'ubkg_code': 'C201090', 'assaytype': 'confocal', 'vitessce-hints': [], 'dir-schema': 'confocal-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'Confocal', 'description': 'Confocal Microscopy'}",
        "rule_description": "DCWG confocal"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Thick section Multiphoton MxIF'",
        "value": "{'ubkg_code': 'C201100', 'assaytype': 'thick-section-multiphoton-mxif', 'vitessce-hints': [], 'dir-schema': 'thick-section-multiphoton-mxif-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'Thick section Multiphoton MxIF', 'description': 'Thick section Multiphoton MxIF'}",
        "rule_description": "DCWG thick-section-multiphoton-mxif"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Second Harmonic Generation (SHG)'",
        "value": "{'ubkg_code': 'C201110', 'assaytype': 'second-harmonic-generation', 'vitessce-hints': [], 'dir-schema': 'second-harmonic-generation-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'Second Harmonic Generation (SHG)', 'description': 'Second Harmonic Generation (SHG)'}",
        "rule_description": "DCWG second-harmonic-generation"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Enhanced Stimulated Raman Spectroscopy (SRS)'",
        "value": "{'ubkg_code': 'C201120', 'assaytype': 'enhanced-srs', 'vitessce-hints': [], 'dir-schema': 'enhanced-srs-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'Enhanced Stimulated Raman Spectroscopy (SRS)', 'description': 'Enhanced Stimulated Raman Spectroscopy (SRS)'}",
        "rule_description": "DCWG enhanced-srs"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Molecular Cartography'",
        "value": "{'ubkg_code': 'C201130', 'assaytype': 'molecular-cartography', 'vitessce-hints': [], 'dir-schema': 'mc-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'Molecular Cartography', 'description': 'Molecular Cartography'}",
        "rule_description": "DCWG molecular-cartography"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['visium_no_probes']",
        "value": "{'ubkg_code': 'C201140', 'assaytype': 'visium-no-probes', 'vitessce-hints': ['rna', 'is_image', 'anndata', 'spatial'], 'primary': false, 'contains-pii': false, 'description': 'Visium (no probes) [Salmon + Scanpy]', 'is-multi-assay': true, 'pipeline-shorthand': 'Salmon + Scanpy'}",
        "rule_description": "derived visium-no-probes"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['multiome_snareseq']",
        "value": "{'ubkg_code': 'C201150', 'assaytype': 'multiome-snare-seq2', 'vitessce-hints': ['rna', 'atac', 'spatial', 'anndata'], 'primary': false, 'contains-pii': false, 'description': 'SNAREseq2 [Salmon + ArchR + Muon]', 'is-multi-assay': true, 'pipeline-shorthand': 'Salmon + ArchR + Muon'}",
        "rule_description": "derived multiome snare-seq2"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['multiome_10x']",
        "value": "{'ubkg_code': 'C201160', 'assaytype': 'multiome-10x', 'vitessce-hints': ['is_sc', 'rna', 'atac', 'anndata'], 'primary': false, 'contains-pii': false, 'description': '10x Multiome [Salmon + ArchR + Muon]', 'is-multi-assay': true, 'pipeline-shorthand': 'Salmon + ArchR + Muon'}",
        "rule_description": "derived multiome 10x"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'MUSIC'",
        "value": "{'ubkg_code': 'C201170', 'assaytype': 'music', 'vitessce-hints': [], 'dir-schema': 'music-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'MUSIC', 'description': 'MUSIC'}",
        "rule_description": "DCWG music"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['phenocycler_deepcell', 'phenocycler_low_cell_count', 'phenocycler_high_cell_count']",
        "value": "{'ubkg_code': 'C201180', 'assaytype': 'phenocycler_deepcell', 'vitessce-hints': ['is_image', 'is_tiled', 'sprm', 'anndata'], 'primary': false, 'contains-pii': false, 'description': 'PhenoCycler [DeepCell + SPRM]', 'pipeline-shorthand': 'DeepCell + SPRM'}",
        "rule_description": "derived phenocycler"
    },
    {
        "type": "match",
        "match": "is_epic and derived_dataset_type == 'Segmentation Mask'",
        "value": "{'ubkg_code': 'C202020', 'assaytype': 'segmentation-mask', 'vitessce-hints': ['segmentation_mask', 'is_image', 'pyramid', 'epic'], 'dir-schema': 'segmentation-mask-v2', 'contains-pii': false, 'primary': false, 'dataset-type': 'Segmentation Mask', 'description': 'Segmentation Mask', 'process_state': 'epic'}",
        "rule_description": "EPIC dataset segmentation mask"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'RNAseq (with probes)' and oligo_probe_panel in ['NanoString Technologies; GeoMx Human Whole Transcriptome Atlas, 4 slides; PN GMX-RNA-NGS-HuWTA-4', 'NanoString Technologies; GeoMx Mouse Whole Transcriptome Atlas, 4 slides; PN GMX-RNA-NGS-MsWTA-4'] and assay_input_entity == 'area of interest'",
        "value": "{'ubkg_code': 'C202030', 'assaytype': 'geomx-rnaseq-with-probes', 'vitessce-hints': [], 'dir-schema': 'rnaseq-with-probes-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'RNAseq (with probes)', 'description': 'GeoMx RNAseq (with probes)'}",
        "rule_description": "DCWG RNAseq-geomx-with-probes"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'RNAseq (with probes)' and oligo_probe_panel in ['10X Genomics; Chromium Next GEM Single Cell Fixed RNA Human Transcriptome Probe Kit, 64 rxns; PN 1000456', '10X Genomics; Chromium Next GEM Single Cell Fixed RNA Human Transcriptome Probe Kit, 16 rxns; PN 1000420', '10X Genomics; Chromium Next GEM Single Cell Fixed RNA, Mouse Transcriptome Probe Kit, 64 rxns; PN 1000492', '10x Genomics; Chromium Fixed RNA Kit, Human Transcriptome 4 rxns x 1 BC; PN 1000474'] and assay_input_entity == 'single cell' and barcode_read =~~ 'Read 1' and barcode_size == 16 and barcode_offset == 0 and umi_read =~~ 'Read 1' and umi_size == 12 and umi_offset == 16",
        "value": "{'ubkg_code': 'C202040', 'assaytype': 'scRNAseq-with-probes', 'vitessce-hints': [], 'dir-schema': 'rnaseq-with-probes-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'RNAseq (with probes)', 'description': 'RNAseq (with probes)'}",
        "rule_description": "DCWG scRNAseq-with-probes"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Xenium' and oligo_probe_panel in ['10x Genomics; Xenium Human Multi-Tissue and Cancer Panel v1; PN 1000626', '10x Genomics; Xenium Prime 5K Human Pan Tissue & Pathways Panel; PN 1000724', 'Custom', '10x Genomics; Xenium Human Colon Gene Expression Panel; PN 1000642', '10x Genomics; Xenium Human Lung Gene Expression Panel; PN 1000601', '10x Genomics; Xenium Custom Gene Expression Panel (up to 50 genes), PN-1000464', '10x Genomics; Xenium Custom Gene Expression Panel, (51-100 genes); PN 1000561', '10x Genomics; Xenium Mouse Multi-Tissue Atlassing Panel; PN 1000627']",
        "value": "{'ubkg_code': 'C202050', 'assaytype': 'xenium', 'vitessce-hints': [], 'dir-schema': 'xenium-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'Xenium', 'description': 'Xenium'}",
        "rule_description": "DCWG xenium"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['kaggle_2_segmentation']",
        "value": "{'ubkg_code': 'C202060', 'assaytype': 'h-and-e', 'vitessce-hints': ['segmentation_mask', 'is_image', 'pyramid'], 'primary': false, 'contains-pii': false, 'description': 'H&E Stained Microscopy [Kaggle-2 Segmentation]', 'pipeline-shorthand': 'Kaggle-2 Segmentation'}",
        "rule_description": "derived h-and-e"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'CyTOF'",
        "value": "{'ubkg_code': 'C202070', 'assaytype': 'cytof', 'vitessce-hints': [], 'dir-schema': 'cytof-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'CyTOF', 'description': 'CyTOF'}",
        "rule_description": "DCWG CyTOF"
    },
    {
      "type": "match",
      "match": "is_dcwg and is_primary and dataset_type == 'Histology'  and stain_name == 'AB-PAS'",
      "value": "{'ubkg_code': 'C202080', 'assaytype': 'AB-PAS', 'vitessce-hints': [], 'dir-schema': 'histology-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'Histology', 'description': 'AB-PAS Histology'}",
      "rule_description": "DCWG AB-PAS Histology"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'DBiT-seq'",
        "value": "{'ubkg_code': 'C202090', 'assaytype': 'dbit-seq', 'vitessce-hints': [], 'dir-schema': 'dbit-seq-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'DBiT-seq', 'description': 'DBiT-seq'}",
        "rule_description": "DCWG DBiT-seq"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Singular Genomics G4X'",
        "value": "{'ubkg_code': 'C202100', 'assaytype': 'singular-genomics-g4x', 'vitessce-hints': [], 'dir-schema': 'singular-genomics-g4x-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'Singular Genomics G4X', 'description': 'Singular Genomics G4X'}",
        "rule_description": "DCWG Singular Genomics G4X"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['geomx']",
        "value": "{'ubkg_code': 'C202110', 'assaytype': 'geomx_ngs', 'vitessce-hints': ['geomx', 'is_image', 'pyramid'], 'primary': false, 'contains-pii': false, 'description': 'GeoMx (NGS) [AnnData]', 'pipeline-shorthand': 'AnnData'}",
        "rule_description": "derived geomx_ngs"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'CosMx Transcriptomics' and oligo_probe_panel in ['NanoString Technologies; CosMx Human Universal Cell Characterization Panel (RNA, 1000 Plex); PN CMX-H-USCP-1KP-R', 'NanoString Technologies; CosMx Hs WTX RNA Panel Kit, 2 slides: PN 121500047', 'NanoString Technologies; CosMx Hs Univ Cell (RNA, 1000 Plex); PN 121500002', 'NanoString Technologies; CosMx Human 6K Discovery Panel (RNA, 6175 Plex); PN 121500041', 'NanoString Technologies; CosMx Mouse Universal Cell Characterization Panel (RNA, 1000 Plex); PN CMX-M-USCP-1KP-R', 'NanoString Technologies; CosMx Mouse Neuroscience Panel (RNA, 1000 Plex); PN CMX-M-NEUP-R']",
        "value": "{'ubkg_code': 'C202120', 'assaytype': 'cosmx_transcriptomics', 'vitessce-hints': [], 'dir-schema': 'cosmx-transcriptomics-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'CosMx Transcriptomics', 'description': 'CosMx Transcriptomics'}",
        "rule_description": "DCWG CosMx Transcriptomics"
    },
    {
        "type": "match",
        "match": "is_central_processed and data_types[0] in ['visium_with_probes']",
        "value": "{'ubkg_code': 'C202130', 'assaytype': 'visium-with-probes', 'vitessce-hints': ['rna', 'is_image', 'anndata', 'spatial'], 'description': 'Visium (with probes) [Salmon + Scanpy]', 'contains-pii': false, 'primary': false, 'pipeline-shorthand': 'Salmon + Scanpy'}",
        "rule_description": "derived visium-with-probes"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Stereo-seq'",
        "value": "{'ubkg_code': 'C202140', 'assaytype': 'stereo-seq', 'vitessce-hints': [], 'dir-schema': 'stereo-seq-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'Stereo-seq', 'description': 'Stereo-seq'}",
        "rule_description": "DCWG Stereo-seq"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'CosMx Proteomics' and oligo_probe_panel in ['NanoString Technologies; CosMx Human Immuno-Oncology Panel (Protein, 64 Plex); PN CMX-H-IOP-64P-P', 'NanoString Technologies; CosMx Mouse Neuroscience Panel (Protein, 64 Plex); PN CMX-M-Neuro-64P-P']",
        "value": "{'ubkg_code': 'C202150', 'assaytype': 'cosmx_proteomics', 'vitessce-hints': [], 'dir-schema': 'cosmx-proteomics-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'CosMx Proteomics', 'description': 'CosMx Proteomics'}",
        "rule_description": "DCWG CosMx Proteomics"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'FACS'",
        "value": "{'ubkg_code': 'C202160', 'assaytype': 'FACS', 'vitessce-hints': [], 'dir-schema': 'facs-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'FACS', 'description': 'FACS'}",
        "rule_description": "DCWG FACS"
    },
    {
         "type": "match",
         "match": "is_central_processed and data_types[0] in ['xenium']",
         "value": "{'ubkg_code': 'C202170', 'assaytype': 'xenium', 'vitessce-hints': ['xenium', 'is_image'], 'description': 'Xenium [SpatialData + Scanpy]', 'contains-pii': false, 'primary': false, 'pipeline-shorthand': 'SpatialData + Scanpy'}",
         "rule_description": "derived xenium"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Olink'",
        "value": "{'ubkg_code': 'C202180', 'assaytype': 'olink', 'vitessce-hints': [], 'dir-schema': 'olink-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'Olink', 'description': 'Olink'}",
        "rule_description": "DCWG Olink"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'SeqScope'",
        "value": "{'ubkg_code': 'C202190', 'assaytype': 'seqscope', 'vitessce-hints': [], 'dir-schema': 'seq-scope-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'SeqScope', 'description': 'SeqScope'}",
        "rule_description": "DCWG SeqScope"
    },
    {
        "type": "match",
        "match": "is_epic and derived_dataset_type == 'Object x Analyte'",
        "value": "{'ubkg_code': 'C202200', 'assaytype': 'object-x-analyte', 'vitessce-hints': [], 'dir-schema': 'object-x-analyte-v2', 'contains-pii': false, 'primary': false, 'dataset-type': 'Object x Analyte', 'description': 'Object x Analyte', 'process_state': 'epic'}",
        "rule_description": "EPIC dataset Object x Analyte"
    },
    {
         "type": "match",
         "match": "is_central_processed and data_types[0] in ['rna_with_probes']",
         "value": "{'ubkg_code': 'C202210', 'assaytype': 'rna_with_probes', 'vitessce-hints': ['rna', 'is_sc'], 'description': 'RNAseq (with probes) [BWA + Scanpy]', 'contains-pii': false, 'primary': false, 'pipeline-shorthand': 'BWA + Scanpy'}",
         "rule_description": "derived RNAseq w/ Probes"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Visium HD'",
        "value": "{'ubkg_code': 'C202220', 'assaytype': 'visium-hd', 'vitessce-hints': [], 'dir-schema': 'visium-hd-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'Visium HD', 'description': 'Visium HD', 'must-contain': ['RNAseq (with probes)','Histology'], 'is-multi-assay': true}",
        "rule_description": "DCWG Visium HD"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'MPLEx'",
        "value": "{'ubkg_code': 'C202230', 'assaytype': 'mplex', 'vitessce-hints': [], 'dir-schema': 'mplex-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'MPLEx', 'description': 'MPLEx'}",
        "rule_description": "DCWG MPLEx"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Pixel-seqV2'",
        "value": "{'ubkg_code': 'C202240', 'assaytype': 'pixel-seqv2', 'vitessce-hints': [], 'dir-schema': 'pixel-seqv2-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'Pixel-seqV2', 'description': 'Pixel-seqV2'}",
        "rule_description": "DCWG Pixel-seqV2"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == '4i'",
        "value": "{'ubkg_code': 'C202250', 'assaytype': '4i', 'vitessce-hints': [], 'dir-schema': '4i-v2', 'contains-pii': false, 'primary': true, 'dataset-type': '4i', 'description': '4i'}",
        "rule_description": "DCWG 4i"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'Raman Imaging'",
        "value": "{'ubkg_code': 'C202260', 'assaytype': 'raman-imaging', 'vitessce-hints': [], 'dir-schema': 'raman-imaging-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'Raman Imaging', 'description': 'Raman Imaging'}",
        "rule_description": "DCWG Raman Imaging"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'STARmap'",
        "value": "{'ubkg_code': 'C202270', 'assaytype': 'starmap', 'vitessce-hints': [], 'dir-schema': 'starmap-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'STARmap', 'description': 'STARmap'}",
        "rule_description": "DCWG STARmap"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'RNAseq (with probes)' and oligo_probe_panel in ['NanoString Technologies; GeoMx Mouse Whole Transcriptome Atlas, 4 slides; PN GMX-RNA-NGS-MsWTA-4', '10X Genomics; GEM-X Flex Human Transcriptome Probe Kit, 16 samples; PN 1000785']",
        "value": "{'ubkg_code': 'C202280', 'assaytype': '10x-flex', 'vitessce-hints': [], 'dir-schema': 'rnaseq-with-probes-v2', 'contains-pii': true, 'primary': true, 'dataset-type': 'RNAseq (with probes)', 'description': '10x Flex'}",
        "rule_description": "DCWG 10x Flex"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'iCLAP'",
        "value": "{'ubkg_code': 'C202290', 'assaytype': 'iclap', 'vitessce-hints': [], 'dir-schema': 'iclap-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'iCLAP', 'description': 'iCLAP'}",
        "rule_description": "DCWG iCLAP"
    },
    {
        "type": "match",
        "match": "is_dcwg and is_primary and dataset_type == 'COMET'",
        "value": "{'ubkg_code': 'C202300', 'assaytype': 'comet', 'vitessce-hints': [], 'dir-schema': 'comet-v2', 'contains-pii': false, 'primary': true, 'dataset-type': 'COMET', 'description': 'COMET'}",
        "rule_description": "DCWG COMET"
    }
  ],
"pre":
  [
    {
        "type": "note",
        "match": "metadata_schema_id == null",
        "value": "{'not_dcwg': true, 'is_dcwg': false}",
        "rule_description": "Preamble rule identifying DCWG"
    },
    {
        "type": "note",
        "match": "metadata_schema_id != null",
        "value": "{'not_dcwg': false, 'is_dcwg': true}",
        "rule_description": "Preamble rule identifying non-DCWG"
    },
    {
        "type": "note",
        "match": "creation_action == 'Create Dataset Activity' or creation_action == 'Multi-Assay Split' or (creation_action == null and (dataset_type != null or assay_type != null))",
        "value": "{'is_primary': true}",
        "rule_description": "Preamble rule identifying primary datasets"
    },
    {
        "type": "note",
        "match": "creation_action == 'Central Process'",
        "value": "{'is_central_processed': true}",
        "rule_description": "Preamble rule identifying centrally processed datasets"
    },
    {
        "type": "note",
        "match": "creation_action == 'Create Publication Activity' or dataset_type == 'Publication' or assay_type == 'publication'",
        "value": "{'is_publication': true}",
        "rule_description": "Preamble rule identifying publications"
    },
    {
        "type": "note",
        "match": "creation_action == 'Lab Process'",
        "value": "{'is_lab_processed': true}",
        "rule_description": "Preamble rule identifying lab processed datasets"
    },
    {
        "type": "note",
        "match": "not_dcwg and is_primary and version == null",
        "value": "{'version': 0}",
        "rule_description": "Cover default schema version = 0 for non-DCWG metadata"
    },
    {
        "type": "note",
        "match": "creation_action == 'External Process' or (parent_dataset_id != null and derived_dataset_type != null)",
        "value": "{'is_epic': true}",
        "rule_description": "Preamble rule identifying EPIC"
    },
    {
        "type": "match",
        "match": "true",
        "value": "{'not_dcwg': not_dcwg, 'is_dcwg': is_dcwg, 'is_primary': is_primary, 'is_central_processed': is_central_processed, 'is_publication': is_publication, 'is_lab_processed': is_lab_processed, 'version': version, 'is_epic': is_epic}",
        "rule_description": "return expected values from preamble"
    }
  ],
"post":
  [
    {
        "type": "note",
        "match": "true",
        "value": "{'final_contains_pii': (source_is_human and (ubkg_values&['contains_full_genetic_sequences'] ? ubkg_values&['contains_full_genetic_sequences'] : values&['contains-pii']))}",
        "rule_description": "non-human data is not PII"
    },
    {
        "type": "note",
        "match": "true",
        "value": "{'final_vitessce_hints': (ubkg_values&['vitessce_hints'] ? ubkg_values&['vitessce_hints'] : values&['vitessce-hints'])}",
        "rule_description": "use UBKG vitessce_hints if available"
    },
    {
        "type": "note",
        "match": "true",
        "value": "{'final_assaytype': (ubkg_values&['assaytype'] ? ubkg_values&['assaytype'] : values&['assaytype'])}",
        "rule_description": "use UBKG assaytype if available"
    },
    {
        "type": "note",
        "match": "true",
        "value": "{'final_dir_schema': (ubkg_values&['dir_schema'] ? ubkg_values&['dir_schema'] : values&['dir-schema'])}",
        "rule_description": "use UBKG dir_schema if available"
    },
    {
        "type": "note",
        "match": "true",
        "value": "{'final_primary': (ubkg_values&['primary'] ? ubkg_values&['primary'] : values&['primary'])}",
        "rule_description": "use UBKG primary if available"
    },
    {
        "type": "note",
        "match": "true",
        "value": "{'final_dataset_type': (ubkg_values&['dataset_type'] ? ubkg_values&['dataset_type'] : values&['dataset-type'])}",
        "rule_description": "use UBKG dataset_type if available"
    },
    {
        "type": "note",
        "match": "true",
        "value": "{'final_description': (ubkg_values&['description'] ? ubkg_values&['description'] : values&['description'])}",
        "rule_description": "use UBKG description if available"
    },
    {
        "type": "note",
        "match": "true",
        "value": "{'final_must_contain': (ubkg_values&['must_contain'] ? ubkg_values&['must_contain'] : values&['must-contain'])}",
        "rule_description": "use UBKG must_contain if available"
    },
    {
        "type": "note",
        "match": "true",
        "value": "{'final_is_multi_assay': (ubkg_values&['is_multi_assay'] ? ubkg_values&['is_multi_assay'] : values&['is-multi-assay'])}",
        "rule_description": "use UBKG must_contain if available"
    },
    {
        "type": "note",
        "match": "true",
        "value": "{'final_pipeline_shorthand': (ubkg_values&['pipeline_shorthand'] ? ubkg_values&['pipeline_shorthand'] : values&['pipeline-shorthand'])}",
        "rule_description": "use UBKG must_contain if available"
    },
    {
        "type": "note",
        "match": "true",
        "value": "{'final_tbl_schema': (ubkg_values&['tbl_schema'] ? ubkg_values&['tbl_schema']+pre_values['version'].to_str : values&['tbl-schema'])}",
        "rule_description": "use UBKG must_contain if available"
    },
    {
        "type": "note",
        "match": "true",
        "value": "{'final_process_state': (ubkg_values&['process_state'] ? ubkg_values&['process_state'] : '')}",
        "rule_description": "use UBKG process_state if available"
    },
    {
        "type": "match",
        "match": "true",
        "value": "{'contains-pii': final_contains_pii, 'vitessce-hints': final_vitessce_hints, 'assaytype': final_assaytype, 'dir-schema': final_dir_schema, 'primary': final_primary, 'dataset-type': final_dataset_type, 'description': final_description, 'must-contain': final_must_contain, 'is-multi-assay': final_is_multi_assay, 'pipeline-shorthand': final_pipeline_shorthand, 'tbl-schema': final_tbl_schema, 'process_state': final_process_state, 'ubkg_code': values&['ubkg_code']}",
        "rule_description": "return expected values"
    }
  ]
}