Merge pull request #741 from microbiomedata/handle-multiple-envs

Add mapping for multiple environments

nmdc_runtime/site/translation/submission_portal_translator.py

@@ -1,21 +1,44 @@
 import logging
 import re
 from datetime import datetime
+from enum import Enum
 from functools import lru_cache
 from importlib import resources
 from typing import Any, List, Optional, Union
 
 from linkml_runtime import SchemaView
 from linkml_runtime.linkml_model import SlotDefinition
 from nmdc_schema import nmdc
-from toolz import get_in, groupby, concat, valmap, dissoc
+from toolz import concat, dissoc, get_in, groupby, valmap
 
 from nmdc_runtime.site.translation.translator import JSON_OBJECT, Translator
 
-
 BIOSAMPLE_UNIQUE_KEY_SLOT = "samp_name"
 
 
+class EnvironmentPackage(Enum):
+    r"""
+    Enumeration of all possible environmental packages.
+
+    >>> EnvironmentPackage.AIR.value
+    'air'
+    >>> EnvironmentPackage.SEDIMENT.value
+    'sediment'
+    """
+
+    AIR = "air"
+    BIOFILM = "microbial mat_biofilm"
+    BUILT_ENV = "built environment"
+    HCR_CORES = "hydrocarbon resources-cores"
+    HRC_FLUID_SWABS = "hydrocarbon resources-fluids_swabs"
+    HOST_ASSOCIATED = "host-associated"
+    MISC_ENVS = "miscellaneous natural or artificial environment"
+    PLANT_ASSOCIATED = "plant-associated"
+    SEDIMENT = "sediment"
+    SOIL = "soil"
+    WATER = "water"
+
+
 @lru_cache
 def _get_schema_view():
     """Return a SchemaView instance representing the NMDC schema"""
@@ -550,7 +573,6 @@ def _translate_biosample(
         sample_data: List[JSON_OBJECT],
         nmdc_biosample_id: str,
         nmdc_study_id: str,
-        default_env_package: str,
     ) -> nmdc.Biosample:
         """Translate sample data from portal submission into an `nmdc:Biosample` object.
 
@@ -565,18 +587,23 @@ def _translate_biosample(
                             from each applicable submission portal tab
         :param nmdc_biosample_id: Minted nmdc:Biosample identifier for the translated object
         :param nmdc_study_id: Minted nmdc:Study identifier for the related Study
-        :param default_env_package: Default value for `env_package` slot
         :return: nmdc:Biosample
         """
-        biosample_key = sample_data[0].get(BIOSAMPLE_UNIQUE_KEY_SLOT, "").strip()
+        env_idx = next(
+            (
+                i
+                for i, tab in enumerate(sample_data)
+                if tab.get("env_package") is not None
+            ),
+            0,
+        )
+        biosample_key = sample_data[env_idx].get(BIOSAMPLE_UNIQUE_KEY_SLOT, "").strip()
         slots = {
             "id": nmdc_biosample_id,
             "associated_studies": [nmdc_study_id],
             "type": "nmdc:Biosample",
-            "name": sample_data[0].get("samp_name", "").strip(),
-            "env_package": nmdc.TextValue(
-                has_raw_value=default_env_package, type="nmdc:TextValue"
-            ),
+            "name": sample_data[env_idx].get("samp_name", "").strip(),
+            "env_package": sample_data[env_idx].get("env_package"),
         }
         for tab in sample_data:
             transformed_tab = self._transform_dict_for_class(tab, "Biosample")
@@ -613,9 +640,18 @@ def get_database(self) -> nmdc.Database:
         ]
 
         sample_data = metadata_submission_data.get("sampleData", {})
-        package_name = metadata_submission_data["packageName"]
+        for key in sample_data.keys():
+            env = key.removesuffix("_data").upper()
+            try:
+                package_name = EnvironmentPackage[env].value
+                for sample in sample_data[key]:
+                    sample["env_package"] = package_name
+            except KeyError:
+                pass
+
         sample_data_by_id = groupby(
-            BIOSAMPLE_UNIQUE_KEY_SLOT, concat(sample_data.values())
+            BIOSAMPLE_UNIQUE_KEY_SLOT,
+            concat(sample_data.values()),
         )
         nmdc_biosample_ids = self._id_minter("nmdc:Biosample", len(sample_data_by_id))
         sample_data_to_nmdc_biosample_ids = dict(
@@ -627,7 +663,6 @@ def get_database(self) -> nmdc.Database:
                 sample_data,
                 nmdc_biosample_id=sample_data_to_nmdc_biosample_ids[sample_data_id],
                 nmdc_study_id=nmdc_study_id,
-                default_env_package=package_name,
             )
             for sample_data_id, sample_data in sample_data_by_id.items()
             if sample_data

tests/test_data/test_submission_portal_translator_data.yaml

@@ -2,7 +2,7 @@
 input:
   metadata_submission:
     metadata_submission:
-      packageName: plant-associated
+      packageName: ['plant-associated', 'air']
       contextForm:
         datasetDoi: doi:10.46936/10.25585/60000818
         dataGenerated: true
@@ -31,8 +31,7 @@ input:
         biosafetyLevel: ''
         irbOrHipaa:
         comments: ''
-      templates:
-      - plant-associated
+      templates: ['plant-associated', 'air', 'jgi_mg']
       studyForm:
         studyName: Seasonal activities of the phyllosphere microbiome of perennial crops
         piName: Ashley Shade
@@ -57,7 +56,7 @@ input:
           annotated. Our findings support that leaf-associated bacterial populations are
           seasonally dynamic and responsive to host cues.
         notes: ''
-        fundingSources: 
+        fundingSources:
           - Some award ABC
           - Some award XYZ
         contributors:
@@ -281,6 +280,36 @@ input:
           ecosystem_subtype: Leaf
           ecosystem_category: Terrestrial
           specific_ecosystem: Phyllosphere
+        air_data:
+        - elev: 225
+          lat_lon: 50.586825 6.408977
+          samp_name: "wedw"
+          env_medium: abcd [ENVO:00001998]
+          geo_loc_name: "USA: Maryland, Bethesda"
+          analysis_type:
+          - metaproteomics
+          collection_date: 2021
+          env_broad_scale: ewdwed [ENVO:123]
+          env_local_scale: asdxasd [ENV:234]
+          samp_store_temp: "-80 Celsius"
+        jgi_mg_data:
+        - samp_name: "G6R2_MAIN_09MAY2016"
+          analysis_type:
+            - metagenomics
+          dna_concentration: 0.1
+          dna_cont_type: tube
+          dna_container_id: 00001
+          dna_dnase: "no"
+          dna_isolate_meth: phenol/chloroform extraction
+          dna_project_contact: Jane Doe
+          dna_samp_id: 00001
+          dna_sample_format: Ethanol
+          dna_sample_name: 00001
+          dna_seq_project: 00001
+          dna_seq_project_name: 00001
+          dna_seq_project_pi: Jane Doe
+          dna_volume: 10
+          proposal_dna: 00001
     status: in-progress
     id: d32b5eb6-71e8-4c98-8a57-3e64d05718b4
     author_orcid: 0000-0002-7705-343X
@@ -933,6 +962,69 @@ output:
       has_raw_value: UUID:c0c4a2b5-0382-450a-8728-a176fa438efe
     analysis_type:
     - metagenomics
+    dna_concentration: 0.1
+    dna_cont_type: tube
+    dna_container_id: 00001
+    dna_dnase: "no"
+    dna_isolate_meth: phenol/chloroform extraction
+    dna_project_contact: Jane Doe
+    dna_samp_id: 00001
+    dna_sample_format: Ethanol
+    dna_sample_name: 00001
+    dna_seq_project: 00001
+    dna_seq_project_name: 00001
+    dna_seq_project_pi: Jane Doe
+    dna_volume: 10
+    proposal_dna: 00001
+  - id: nmdc:bsm-00-7r7pn0r2
+    type: 'nmdc:Biosample'
+    name: 'wedw'
+    associated_studies:
+      - 'nmdc:sty-00-y0cq65zt'
+    env_broad_scale:
+      type: 'nmdc:ControlledIdentifiedTermValue'
+      has_raw_value: 'ewdwed [ENVO:123]'
+      term:
+        id: 'ENVO:123'
+        type: 'nmdc:OntologyClass'
+        name: 'ewdwed'
+    env_local_scale:
+      type: 'nmdc:ControlledIdentifiedTermValue'
+      has_raw_value: 'asdxasd [ENV:234]'
+      term:
+        id: 'ENV:234'
+        type: 'nmdc:OntologyClass'
+        name: 'asdxasd'
+    env_medium:
+      type: 'nmdc:ControlledIdentifiedTermValue'
+      has_raw_value: 'abcd [ENVO:00001998]'
+      term:
+        id: 'ENVO:00001998'
+        type: 'nmdc:OntologyClass'
+        name: 'abcd'
+    samp_name: 'wedw'
+    collection_date:
+      type: 'nmdc:TimestampValue'
+      has_raw_value: '2021'
+    elev: 225.0
+    env_package:
+      type: 'nmdc:TextValue'
+      has_raw_value: 'air'
+    geo_loc_name:
+      type: 'nmdc:TextValue'
+      has_raw_value: 'USA: Maryland, Bethesda'
+    lat_lon:
+      type: 'nmdc:GeolocationValue'
+      has_raw_value: '50.586825 6.408977'
+      latitude: 50.586825
+      longitude: 6.408977
+    samp_store_temp:
+      type: 'nmdc:QuantityValue'
+      has_raw_value: '-80 Celsius'
+      has_unit: 'Celsius'
+      has_numeric_value: -80.0
+    analysis_type:
+    - metaproteomics
   study_set:
   - id: nmdc:sty-00-y0cq65zt
     type: nmdc:Study
@@ -1032,7 +1124,7 @@ input:
             roles:
               - Principal Investigator
         description: This is a test submission
-        fundingSources: 
+        fundingSources:
           - Some award ABC
           - Some award XYZ
         linkOutWebpage: