{
"domain": "biotech",
"entities": [
{
"name": "laboratory",
"weight": 1,
"columns": [
{
"name": "lab_id",
"type": "string"
},
{
"name": "lab_name",
"type": "string",
"pool_name": "company"
},
{
"name": "lab_type",
"type": "string",
"pool_name": "word"
},
{
"name": "director_name",
"type": "string",
"pool_name": "first_name"
},
{
"name": "city",
"type": "string"
},
{
"name": "country_code",
"type": "string"
},
{
"name": "accreditation",
"type": "string",
"pool_name": "word",
"null_rate_default": 0.3
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": []
},
{
"name": "research_project",
"weight": 10,
"columns": [
{
"name": "project_id",
"type": "string"
},
{
"name": "lab_id",
"type": "string"
},
{
"name": "project_code",
"type": "string"
},
{
"name": "project_title",
"type": "string",
"pool_name": "word"
},
{
"name": "principal_investigator",
"type": "string",
"pool_name": "first_name"
},
{
"name": "grant_amount",
"type": "float"
},
{
"name": "start_date",
"type": "date"
},
{
"name": "end_date",
"type": "date",
"null_rate_default": 0.25
},
{
"name": "status",
"type": "string",
"pool_name": "status_active"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": [
{
"source_col": "lab_id",
"target_entity": "laboratory"
}
]
},
{
"name": "sample",
"weight": 50,
"columns": [
{
"name": "sample_id",
"type": "string"
},
{
"name": "project_id",
"type": "string"
},
{
"name": "sample_barcode",
"type": "string"
},
{
"name": "sample_type",
"type": "string",
"pool_name": "sample_type"
},
{
"name": "collection_date",
"type": "date"
},
{
"name": "collected_by",
"type": "string",
"pool_name": "first_name"
},
{
"name": "storage_condition",
"type": "string",
"pool_name": "word"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": [
{
"source_col": "project_id",
"target_entity": "research_project"
}
]
},
{
"name": "experiment",
"weight": 100,
"columns": [
{
"name": "experiment_id",
"type": "string"
},
{
"name": "sample_id",
"type": "string"
},
{
"name": "experiment_type",
"type": "string",
"pool_name": "experiment_type"
},
{
"name": "performed_by",
"type": "string",
"pool_name": "first_name"
},
{
"name": "experiment_date",
"type": "date"
},
{
"name": "result",
"type": "string",
"pool_name": "qc_result"
},
{
"name": "instrument_used",
"type": "string",
"pool_name": "biotech_equipment"
},
{
"name": "notes",
"type": "string",
"pool_name": "word",
"null_rate_default": 0.4
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": [
{
"source_col": "sample_id",
"target_entity": "sample"
}
]
}
],
"hn_types": [
{
"name": "same_project_code_diff_lab",
"entity_type": "research_project",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"project_code\"], \"attr_fields\": [\"project_title\", \"principal_investigator\", \"grant_amount\"]}"
},
{
"name": "same_pi_same_lab",
"entity_type": "research_project",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"principal_investigator\", \"lab_id\"], \"attr_fields\": [\"project_code\", \"grant_amount\", \"status\"]}"
},
{
"name": "same_barcode_diff_project",
"entity_type": "sample",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"sample_barcode\"], \"attr_fields\": [\"sample_type\", \"collection_date\", \"collected_by\"]}"
},
{
"name": "same_experiment_type_same_sample",
"entity_type": "experiment",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"experiment_type\", \"sample_id\"], \"attr_fields\": [\"performed_by\", \"result\", \"instrument_used\"]}"
},
{
"name": "same_instrument_same_date",
"entity_type": "experiment",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"instrument_used\", \"experiment_date\"], \"attr_fields\": [\"experiment_type\", \"sample_id\", \"result\"]}"
}
]
}