for release

nmdc_schema/nmdc.py

@@ -1,5 +1,5 @@
 # Auto generated from nmdc.yaml by pythongen.py version: 0.0.1
-# Generation date: 2024-07-24T13:21:38
+# Generation date: 2024-08-21T17:07:04
 # Schema: NMDC
 #
 # id: https://w3id.org/nmdc/nmdc
@@ -71,6 +71,7 @@
 PANTHER_FAMILY = CurieNamespace('PANTHER_FAMILY', 'https://bioregistry.io/panther.family:')
 PATO = CurieNamespace('PATO', 'http://purl.obolibrary.org/obo/PATO_')
 PFAM = CurieNamespace('PFAM', 'https://bioregistry.io/pfam:')
+PFAM_CLAN = CurieNamespace('PFAM_CLAN', 'https://bioregistry.io/pfam.clan:')
 PO = CurieNamespace('PO', 'http://purl.obolibrary.org/obo/PO_')
 PR = CurieNamespace('PR', 'http://purl.obolibrary.org/obo/PR_')
 PUBCHEM_COMPOUND = CurieNamespace('PUBCHEM_COMPOUND', 'https://bioregistry.io/pubchem.compound:')
@@ -330,6 +331,39 @@ class NomAnalysisActivityId(WorkflowExecutionActivityId):
     pass
 
 
+@dataclass
+class EukEval(YAMLRoot):
+    """
+    This class contains information pertaining to evaluating if a Metagenome-Assembled Genome (MAG) is eukaryotic.
+    """
+    _inherited_slots: ClassVar[List[str]] = []
+
+    class_class_uri: ClassVar[URIRef] = NMDC["EukEval"]
+    class_class_curie: ClassVar[str] = "nmdc:EukEval"
+    class_name: ClassVar[str] = "EukEval"
+    class_model_uri: ClassVar[URIRef] = NMDC.EukEval
+
+    completeness: Optional[float] = None
+    contamination: Optional[float] = None
+    ncbi_lineage_tax_ids: Optional[str] = None
+    ncbi_lineage: Optional[str] = None
+
+    def __post_init__(self, *_: List[str], **kwargs: Dict[str, Any]):
+        if self.completeness is not None and not isinstance(self.completeness, float):
+            self.completeness = float(self.completeness)
+
+        if self.contamination is not None and not isinstance(self.contamination, float):
+            self.contamination = float(self.contamination)
+
+        if self.ncbi_lineage_tax_ids is not None and not isinstance(self.ncbi_lineage_tax_ids, str):
+            self.ncbi_lineage_tax_ids = str(self.ncbi_lineage_tax_ids)
+
+        if self.ncbi_lineage is not None and not isinstance(self.ncbi_lineage, str):
+            self.ncbi_lineage = str(self.ncbi_lineage)
+
+        super().__post_init__(**kwargs)
+
+
 @dataclass
 class FailureCategorization(YAMLRoot):
     _inherited_slots: ClassVar[List[str]] = []
@@ -4721,6 +4755,7 @@ class MagBin(YAMLRoot):
     number_of_contig: Optional[int] = None
     total_bases: Optional[int] = None
     type: Optional[str] = None
+    eukaryotic_evaluation: Optional[Union[dict, EukEval]] = None
 
     def __post_init__(self, *_: List[str], **kwargs: Dict[str, Any]):
         if self.bin_name is not None and not isinstance(self.bin_name, str):
@@ -4783,6 +4818,9 @@ def __post_init__(self, *_: List[str], **kwargs: Dict[str, Any]):
         if self.type is not None and not isinstance(self.type, str):
             self.type = str(self.type)
 
+        if self.eukaryotic_evaluation is not None and not isinstance(self.eukaryotic_evaluation, EukEval):
+            self.eukaryotic_evaluation = EukEval(**as_dict(self.eukaryotic_evaluation))
+
         super().__post_init__(**kwargs)
 
 
@@ -8747,6 +8785,16 @@ class ProcessingInstitutionEnum(EnumDefinitionImpl):
 class slots:
     pass
 
+slots.eukaryotic_evaluation = Slot(uri=NMDC.eukaryotic_evaluation, name="eukaryotic_evaluation", curie=NMDC.curie('eukaryotic_evaluation'),
+                   model_uri=NMDC.eukaryotic_evaluation, domain=None, range=Optional[Union[dict, EukEval]])
+
+slots.ncbi_lineage_tax_ids = Slot(uri=NMDC.ncbi_lineage_tax_ids, name="ncbi_lineage_tax_ids", curie=NMDC.curie('ncbi_lineage_tax_ids'),
+                   model_uri=NMDC.ncbi_lineage_tax_ids, domain=None, range=Optional[str],
+                   pattern=re.compile(r'^\d+(-\d+)*$'))
+
+slots.ncbi_lineage = Slot(uri=NMDC.ncbi_lineage, name="ncbi_lineage", curie=NMDC.curie('ncbi_lineage'),
+                   model_uri=NMDC.ncbi_lineage, domain=None, range=Optional[str])
+
 slots.has_failure_categorization = Slot(uri=NMDC.has_failure_categorization, name="has_failure_categorization", curie=NMDC.curie('has_failure_categorization'),
                    model_uri=NMDC.has_failure_categorization, domain=None, range=Optional[Union[Union[dict, FailureCategorization], List[Union[dict, FailureCategorization]]]])
 

nmdc_schema/nmdc.schema.json

@@ -4103,6 +4103,29 @@
             "title": "EnvironmentalMaterialTerm",
             "type": "object"
         },
+        "EukEval": {
+            "additionalProperties": false,
+            "description": "This class contains information pertaining to evaluating if a Metagenome-Assembled Genome (MAG) is eukaryotic.",
+            "properties": {
+                "completeness": {
+                    "type": "number"
+                },
+                "contamination": {
+                    "type": "number"
+                },
+                "ncbi_lineage": {
+                    "description": "Comma delimited ordered list of NCBI taxonomy names.",
+                    "type": "string"
+                },
+                "ncbi_lineage_tax_ids": {
+                    "description": "Dash-delimited ordered list of NCBI taxonomy identifiers (TaxId)",
+                    "pattern": "^\\d+(-\\d+)*$",
+                    "type": "string"
+                }
+            },
+            "title": "EukEval",
+            "type": "object"
+        },
         "ExtWallOrientEnum": {
             "description": "",
             "enum": [
@@ -5258,6 +5281,10 @@
                 "contamination": {
                     "type": "number"
                 },
+                "eukaryotic_evaluation": {
+                    "$ref": "#/$defs/EukEval",
+                    "description": "Contains results from evaluating if a Metagenome-Assembled Genome is of eukaryotic lineage."
+                },
                 "gene_count": {
                     "type": "integer"
                 },

nmdc_schema/nmdc_materialized_patterns.schema.json

@@ -4112,6 +4112,29 @@
             "title": "EnvironmentalMaterialTerm",
             "type": "object"
         },
+        "EukEval": {
+            "additionalProperties": false,
+            "description": "This class contains information pertaining to evaluating if a Metagenome-Assembled Genome (MAG) is eukaryotic.",
+            "properties": {
+                "completeness": {
+                    "type": "number"
+                },
+                "contamination": {
+                    "type": "number"
+                },
+                "ncbi_lineage": {
+                    "description": "Comma delimited ordered list of NCBI taxonomy names.",
+                    "type": "string"
+                },
+                "ncbi_lineage_tax_ids": {
+                    "description": "Dash-delimited ordered list of NCBI taxonomy identifiers (TaxId)",
+                    "pattern": "^\\d+(-\\d+)*$",
+                    "type": "string"
+                }
+            },
+            "title": "EukEval",
+            "type": "object"
+        },
         "ExtWallOrientEnum": {
             "description": "",
             "enum": [
@@ -5276,6 +5299,10 @@
                 "contamination": {
                     "type": "number"
                 },
+                "eukaryotic_evaluation": {
+                    "$ref": "#/$defs/EukEval",
+                    "description": "Contains results from evaluating if a Metagenome-Assembled Genome is of eukaryotic lineage."
+                },
                 "gene_count": {
                     "type": "integer"
                 },

nmdc_schema/nmdc_materialized_patterns.yaml

@@ -89,6 +89,9 @@ prefixes:
   PFAM:
     prefix_prefix: PFAM
     prefix_reference: 'https://bioregistry.io/pfam:'
+  PFAM.CLAN:
+    prefix_prefix: PFAM.CLAN
+    prefix_reference: 'https://bioregistry.io/pfam.clan:'
   PO:
     prefix_prefix: PO
     prefix_reference: http://purl.obolibrary.org/obo/PO_
@@ -3713,6 +3716,27 @@ enums:
         meaning: https://genomecenter.ucdavis.edu/
         title: University of California, Davis Genome Center
 slots:
+  eukaryotic_evaluation:
+    name: eukaryotic_evaluation
+    description: Contains results from evaluating if a Metagenome-Assembled Genome
+      is of eukaryotic lineage.
+    from_schema: https://w3id.org/nmdc/nmdc
+    range: EukEval
+  ncbi_lineage_tax_ids:
+    name: ncbi_lineage_tax_ids
+    description: Dash-delimited ordered list of NCBI taxonomy identifiers (TaxId)
+    comments:
+    - Example 1-131567-2759-2611352-33682-191814-2603949
+    from_schema: https://w3id.org/nmdc/nmdc
+    range: string
+    pattern: ^\d+(-\d+)*$
+  ncbi_lineage:
+    name: ncbi_lineage
+    description: Comma delimited ordered list of NCBI taxonomy names.
+    comments:
+    - Example root,cellular organisms,Eukaryota,Discoba,Euglenozoa,Diplonemea,Diplonemidae
+    from_schema: https://w3id.org/nmdc/nmdc
+    range: string
   has_failure_categorization:
     name: has_failure_categorization
     from_schema: https://w3id.org/nmdc/nmdc
@@ -18219,6 +18243,19 @@ slots:
     domain: Activity
     range: string
 classes:
+  EukEval:
+    name: EukEval
+    description: This class contains information pertaining to evaluating if a Metagenome-Assembled
+      Genome (MAG) is eukaryotic.
+    comments:
+    - A tool like eukCC (https://doi.org/10.1186/s13059-020-02155-4) would generate
+      information for this class.
+    from_schema: https://w3id.org/nmdc/nmdc
+    slots:
+    - completeness
+    - contamination
+    - ncbi_lineage_tax_ids
+    - ncbi_lineage
   FailureCategorization:
     name: FailureCategorization
     from_schema: https://w3id.org/nmdc/nmdc
@@ -20287,6 +20324,7 @@ classes:
     - number_of_contig
     - total_bases
     - type
+    - eukaryotic_evaluation
   MetaboliteQuantification:
     name: MetaboliteQuantification
     description: This is used to link a metabolomics analysis workflow to a specific