metameq 2026.1.1__py3-none-any.whl

metameq/config/standards.yml

@@ -0,0 +1,1648 @@
+ebi_null_vals_all: &ebi_null_vals_all
+  allowed:
+    - "not collected" 
+    - "not provided"
+    - "restricted access"
+    - "not applicable"
+  type: string
+taxon_id_formats: &taxon_id_formats
+  - regex: ^\d+$
+    type: string
+  - *ebi_null_vals_all
+host_type_specific_metadata:
+  base:
+    metadata_fields:
+      sample_name:
+        empty: false
+        is_phi: false
+        regex: ^[a-zA-Z0-9\.]+$
+        required: true
+        type: string
+        unique: true
+      sample_type:
+        empty: false
+        is_phi: false
+        required: true
+        type: string
+    host_type_specific_metadata:
+      ebi_submittable:
+        "default": "not provided"
+        metadata_fields:
+          anonymized_name:
+            empty: true
+            is_phi: false
+            required: false
+            type: string
+          collection_device:
+            empty: false
+            is_phi: false
+            required: false
+            type: string
+          collection_method:
+            empty: false
+            is_phi: false
+            required: false
+            type: string
+          collection_timestamp:
+            anyof:
+              - regex: '^([0-9]{4})(?:-([0-1][0-9])(?:-([0-3][0-9])(?: ([0-2][0-9])(?::([0-5][0-9])(?::([0-5][0-9]))?)?)?)?)?$'
+                type: string
+                check_with: date_not_in_future
+              - *ebi_null_vals_all
+            empty: false
+            field_desc: The day and time of sampling as a single point in time expressed in
+              24-hour time format, e.g. 2016-11-22.
+            is_phi: false
+            required: true
+          collection_date:
+            anyof:
+              - regex: '^([0-9]{4})(?:-([0-1][0-9])(?:-([0-3][0-9])(?: ([0-2][0-9])(?::([0-5][0-9])(?::([0-5][0-9]))?)?)?)?)?$'
+                type: string
+                check_with: date_not_in_future
+              - *ebi_null_vals_all
+            empty: false
+            field_desc: The day and time of sampling as a single point in time expressed in
+              24-hour time format, e.g. 2016-11-22.
+            is_phi: false
+            required: true
+          country:
+            empty: false
+            is_phi: false
+            required: true
+            type: string
+          description:
+            empty: false
+            field_desc: A description of the sample that can include site, subject, and sample
+              material.
+            is_phi: false
+            required: true
+            type: string
+          dna_extracted:
+            allowed:
+              - 'TRUE'
+              - 'FALSE'
+              - not collected
+              - not provided
+            type: string
+            empty: false
+            field_desc: Whether the DNA been extracted from the sample.
+            is_phi: false
+            required: true
+          elevation:
+            anyof:
+              - min: -413.0
+                type: number
+              - allowed:
+                - not collected
+                - not provided
+                - restricted access
+                type: string
+            empty: false
+            field_desc: Height of land above sea level in meters at the sampling site
+            is_phi: false
+            required: true
+            units: elevation_units
+          elevation_units:
+            allowed:
+              - meters
+            # default: meters
+            empty: false
+            is_phi: false
+            required: false
+            type: string
+          empo_1:
+            empty: false
+            is_phi: false
+            required: true
+            type: string
+          empo_2:
+            empty: false
+            is_phi: false
+            required: true
+            type: string
+          empo_3:
+            empty: false
+            is_phi: false
+            required: true
+            type: string
+          empo_4:
+            # NOT putting the full list of options here: they should be
+            # resolved at the leaf levels
+            empty: false
+            is_phi: false
+            required: true
+            type: string
+          env_biome:
+            empty: false
+            field_desc: Classification of the location where the sample was obtained, from the
+              Environmental Ontology (ENVO).
+            is_phi: false
+            required: true
+            type: string
+          env_feature:
+            empty: false
+            field_desc: Classification of a specific feature in the biome, from the Environmental
+              Ontology (ENVO).
+            is_phi: false
+            required: true
+            type: string
+          env_material:
+            empty: false
+            field_desc: Classification of the material being sampled, from the Environmental
+              Ontology (ENVO).
+            is_phi: false
+            required: true
+            type: string
+          env_package:
+            empty: false
+            field_desc: Environment where the sample was obtained.
+            is_phi: false
+            required: true
+            type: string
+          geo_loc_name:
+            empty: false
+            field_desc: The geographical origin of the sample as defined by the country or location
+              as chosen from the GAZ ontology, e.g. USA:CA:San Diego .
+            is_phi: false
+            required: true
+            type: string
+          latitude:
+            anyof:
+              - max: 90.00
+                min: -90.00
+                type: number
+              - allowed:
+                - not collected
+                - not provided
+                - restricted access
+                type: string
+            empty: false
+            field_desc: Location of the sample site by latitude in decimal degrees, e.g. 32.72
+              for San Diego, USA.
+            is_phi: false
+            required: true
+            units: latitude_units
+          latitude_units:
+            allowed:
+              - Decimal degrees
+            # default: Decimal degrees
+            empty: false
+            is_phi: false
+            required: false
+            type: string
+          longitude:
+            anyof:
+              - max: 180.00
+                min: -180.00
+                type: number
+              - allowed:
+                - not collected
+                - not provided
+                - restricted access
+                type: string
+            empty: false
+            field_desc: Location of the sample site by longitude in decimal degrees, e.g. -117.16
+              for San Diego, USA.
+            is_phi: false
+            required: true
+            units: longitude_units
+          longitude_units:
+            allowed:
+              - Decimal degrees
+            # default: Decimal degrees
+            empty: false
+            is_phi: false
+            required: false
+            type: string
+          physical_specimen_location:
+            empty: false
+            field_desc: Where the sample is stored (if there is any left).
+            is_phi: false
+            required: true
+            type: string
+          physical_specimen_remaining:
+            allowed:
+              - 'TRUE'
+              - 'FALSE'
+              - not collected
+              - not provided
+              - restricted access
+            type: string
+            empty: false
+            field_desc: Whether there is any raw specimen left to sample.
+            is_phi: false
+            required: true
+          scientific_name:
+            empty: false
+            field_desc: The scientific name of the metagenome associated with this sample, as
+              defined in the NCBI Taxonomy. Be aware that this is NOT the scientific name of
+              the host! Note that this value must correspond to the taxon id entered in the
+              taxon_id field; for example, for a sample from the human gut, the scientific_name
+              should be human gut microbiome, corresponding to taxon id 408170.
+            is_phi: false
+            required: true
+            type: string
+          taxon_id:
+            anyof: *taxon_id_formats
+            empty: false
+            field_desc: The NCBI Taxonomy id of the metagenome associated with this sample.
+              Be aware that this is NOT the taxon id of the host! Note that this value must
+              correspond to the scientific name entered in the scientific_name field; for example,
+              for a sample from the human gut, the taxon_id should be 408170, corresponding
+              to the scientific name human gut metagenome.
+            is_phi: false
+            required: true
+          title:
+            empty: false
+            field_desc: Title of study.
+            is_phi: false
+            required: true
+            type: string
+          # TODO: Q: is this field actually required or no?
+#          tube_id:
+#            empty: false
+#            field_desc: The identification of the tube containing the sample.
+#            is_phi: false
+#            required: true
+#            type: string
+        host_type_specific_metadata:
+          host_associated:
+            metadata_fields:
+              empo_1:
+                allowed:
+                  - Host-associated
+                default: Host-associated
+                type: string
+              env_package:
+                allowed:
+                  - host-associated
+                default: host-associated
+                type: string
+              host_age:
+                anyof:
+                  - min_exclusive: 0
+                    type: number
+                  - allowed:
+                    - not collected
+                    - not provided
+                    type: string
+                empty: false
+                field_desc: Host age in appropriate units
+                is_phi: false
+                required: true
+                units: host_age_units
+              host_age_units:
+                empty: false
+                is_phi: false
+                required: true
+                type: string
+              host_common_name:
+                empty: false
+                field_desc: common usage term for host e.g. human
+                is_phi: false
+                required: true
+                type: string
+              host_scientific_name:
+                empty: false
+                field_desc: host scientific name e.g. Homo sapiens
+                is_phi: false
+                required: true
+                type: string
+              host_subject_id:
+                empty: false
+                field_desc: A label that applies to all samples belonging to one subject or one
+                  sample
+                is_phi: false
+                required: true
+                type: string
+              host_taxid:
+                anyof: *taxon_id_formats
+                empty: false
+                field_desc: taxonomy id of host e.g. human is 9606
+                is_phi: false
+                required: true
+              host_weight:
+                anyof:
+                  - min_exclusive: 0
+                    type: number
+                  - allowed:
+                    - not collected
+                    - not provided
+                    type: string
+                empty: false
+                field_desc: in kilograms preferred
+                is_phi: false
+                required: true
+                units: host_weight_units
+              host_weight_units:
+                empty: false
+                is_phi: false
+                required: true
+                type: string
+              # is there any reason time_point has to be host associated??
+              time_point:
+                anyof:
+                  - type: integer
+                  - allowed:
+                    - not applicable
+                    - not collected
+                    - not provided
+                    - restricted access
+                    type: string
+                empty: false
+                field_desc: A numeric ID to indicate order of linked samples
+                is_phi: false
+                required: false
+            host_type_specific_metadata:
+              host_associated_non_saline:
+                metadata_fields:
+                  empo_2:
+                    allowed:
+                      - Host-associated (non-saline)
+                    default: Host-associated (non-saline)
+                    type: string
+                host_type_specific_metadata:
+                  animal_non_saline:
+                    metadata_fields:
+                      empo_3:
+                        allowed:
+                          - Animal (non-saline)
+                        default: Animal (non-saline)
+                        type: string
+                      env_feature:
+                        allowed:
+                          - animal-associated habitat
+                        default: animal-associated habitat
+                        type: string
+                      # apparently uberon ontology is just for animal species..
+                      host_body_habitat:
+                        empty: false
+                        field_desc: from UBERON e.g skin, gastrointestinal tract
+                        is_phi: false
+                        required: true
+                        type: string
+                      host_body_product:
+                        empty: false
+                        field_desc: from UBERON e.g. skin, feces
+                        is_phi: false
+                        required: true
+                        type: string
+                      host_body_site:
+                        empty: false
+                        field_desc: from UBERON e.g. skin of arm, colon
+                        is_phi: false
+                        required: true
+                        type: string
+                      host_life_stage:
+                          # TODO: Gail's writeup includes infant and juvenile
+                          #  but NOT child ... but NPH uses child?
+                        allowed:
+                          - adult
+                          - child
+                          - not applicable
+                          - not collected
+                          - not provided
+                          - restricted access
+                        type: string
+                        empty: false
+                        field_desc: e.g. adult
+                        is_phi: false
+                        required: true
+                      sex:
+                        allowed:
+                          - female
+                          - male
+                          - neuter
+                          - hermaphrodite
+                          - not applicable
+                          - not collected
+                          - not provided
+                          - restricted access
+                        type: string
+                        empty: false
+                        is_phi: false
+                        required: true
+                    host_type_specific_metadata:
+                      vertebrate:
+                        sample_type_specific_metadata:
+                          colon content:
+                            metadata_fields:
+                              env_material:
+                                allowed:
+                                  - intestine environment
+                                default: intestine environment
+                                type: string
+                              host_body_habitat:
+                                allowed:
+                                  - UBERON:gastrointestinal system
+                                default: UBERON:gastrointestinal system
+                                type: string
+                              host_body_product:
+                                allowed:
+                                  - UBERON:lumen of gut
+                                default: UBERON:lumen of gut
+                                type: string
+                              host_body_site:
+                                allowed:
+                                  - UBERON:colon
+                                default: UBERON:colon
+                                type: string
+                              scientific_name:
+                                allowed:
+                                  - gut metagenome
+                                default: gut metagenome
+                                type: string
+                              taxon_id:
+                                allowed:
+                                  - 749906
+                                default: 749906
+                                type: integer
+                          colon mucosa:
+                            metadata_fields:
+                              env_material:
+                                allowed:
+                                  - intestine environment
+                                default: intestine environment
+                                type: string
+                              host_body_habitat:
+                                allowed:
+                                  - UBERON:gastrointestinal system
+                                default: UBERON:gastrointestinal system
+                                type: string
+                              host_body_product:
+                                allowed:
+                                  - UBERON:colon mucosa
+                                default: UBERON:colon mucosa
+                                type: string
+                              host_body_site:
+                                allowed:
+                                  - UBERON:colon
+                                default: UBERON:colon
+                                type: string
+                              scientific_name:
+                                allowed:
+                                  - gut metagenome
+                                default: gut metagenome
+                                type: string
+                              taxon_id:
+                                allowed:
+                                  - 749906
+                                default: 749906
+                                type: integer
+                          saliva:
+                            metadata_fields:
+                              empo_4:
+                                allowed:
+                                  - Animal secretion (non-saline)
+                                default: Animal secretion (non-saline)
+                                type: string
+                              env_material:
+                                allowed:
+                                  - saliva material
+                                default: saliva material
+                                type: string
+                              host_body_habitat:
+                                allowed:
+                                  - UBERON:oral cavity
+                                default: UBERON:oral cavity
+                                type: string
+                              host_body_product:
+                                allowed:
+                                  - UBERON:saliva
+                                default: UBERON:saliva
+                                type: string
+                              host_body_site:
+                                allowed:
+                                  - UBERON:mouth
+                                default: UBERON:mouth
+                                type: string
+                              scientific_name:
+                                allowed:
+                                  - oral metagenome
+                                default: oral metagenome
+                                type: string
+                              taxon_id:
+                                allowed:
+                                  - 1227552
+                                default: 1227552
+                                type: integer
+                          serum:
+                            metadata_fields:
+                              env_material:
+                                allowed:
+                                  - blood
+                                default: blood
+                                type: string
+                              host_body_habitat:
+                                allowed:
+                                  - UBERON:blood
+                                default: UBERON:blood
+                                type: string
+                              host_body_product:
+                                allowed:
+                                  - UBERON:serum
+                                default: UBERON:serum
+                                type: string
+                              host_body_site:
+                                allowed:
+                                  - UBERON:blood plasma
+                                default: UBERON:blood plasma
+                                type: string
+                              scientific_name:
+                                allowed:
+                                  - blood metagenome
+                                default: blood metagenome
+                                type: string
+                              taxon_id:
+                                allowed:
+                                  - 1676987
+                                default: 1676987
+                                type: integer
+                          skin:
+                            metadata_fields:
+                              empo_4:
+                                allowed:
+                                  - Animal surface (non-saline)
+                                default: Animal surface (non-saline)
+                                type: string
+                              env_material:
+                                allowed:
+                                  - sebum material
+                                default: sebum material
+                                type: string
+                              host_body_habitat:
+                                allowed:
+                                  - UBERON:skin
+                                default: UBERON:skin
+                                type: string
+                              host_body_product:
+                                allowed:
+                                  - UBERON:sebum
+                                default: UBERON:sebum
+                                type: string
+                              scientific_name:
+                                allowed:
+                                  - skin metagenome
+                                default: skin metagenome
+                                type: string
+                              taxon_id:
+                                allowed:
+                                  - 1338477
+                                default: 1338477
+                                type: integer
+                          sputum:
+                            metadata_fields:
+                              env_material:
+                                allowed:
+                                  - bodily fluid material
+                                default: bodily fluid material
+                                type: string
+                              host_body_habitat:
+                                allowed:
+                                  - UBERON:respiratory system
+                                default: UBERON:respiratory system
+                                type: string
+                              host_body_product:
+                                allowed:
+                                  - UBERON:sputum
+                                default: UBERON:sputum
+                                type: string
+                              host_body_site:
+                                allowed:
+                                  - UBERON:lung
+                                default: UBERON:lung
+                                type: string
+                              scientific_name:
+                                allowed:
+                                  - lung metagenome
+                                default: lung metagenome
+                                type: string
+                              taxon_id:
+                                allowed:
+                                  - 1729361
+                                default: 1729361
+                                type: integer
+                          # TODO: is this the correct level to introduce feces?
+                          feces:
+                            metadata_fields:
+                              empo_4:
+                                allowed:
+                                  - Animal distal gut (non-saline)
+                                default: Animal distal gut (non-saline)
+                                type: string
+                              env_material:
+                                allowed:
+                                  - fecal material
+                                default: fecal material
+                                type: string
+                              host_body_habitat:
+                                allowed:
+                                  - UBERON:feces
+                                default: UBERON:feces
+                                type: string
+                              host_body_product:
+                                allowed:
+                                  - UBERON:feces
+                                default: UBERON:feces
+                                type: string
+                              host_body_site:
+                                allowed:
+                                  - UBERON:feces
+                                default: UBERON:feces
+                                type: string
+                              scientific_name:
+                                allowed:
+                                  - feces metagenome
+                                default: feces metagenome
+                                type: string
+                              taxon_id:
+                                allowed:
+                                  - 1861841
+                                default: 1861841
+                                type: integer
+                          tongue:
+                            alias: saliva
+                          urine:
+                            metadata_fields:
+                              env_material:
+                                allowed:
+                                  - urine material
+                                default: urine material
+                                type: string
+                              host_body_habitat:
+                                allowed:
+                                  - UBERON:urine
+                                default: UBERON:urine
+                                type: string
+                              host_body_product:
+                                allowed:
+                                  - UBERON:urine
+                                default: UBERON:urine
+                                type: string
+                              host_body_site:
+                                allowed:
+                                  - UBERON:urine
+                                default: UBERON:urine
+                              scientific_name:
+                                allowed:
+                                  - urine metagenome
+                                default: urine metagenome
+                                type: string
+                              taxon_id:
+                                allowed:
+                                  - 1346744
+                                default: 1346744
+                                type: integer
+                        host_type_specific_metadata:
+                          human:
+                            metadata_fields:
+                              env_biome:
+                                # is urban really the only biome for humans?
+                                allowed:
+                                  - urban biome
+                                default: urban biome
+                                type: string
+                              env_feature:
+                                allowed:
+                                  - human-associated habitat
+                                default: human-associated habitat
+                                type: string
+                              env_package:
+                                allowed:
+                                  - human-associated
+                                default: human-associated
+                                type: string
+                              host_age_units:
+                                allowed:
+                                  - years
+                                default: years
+                                type: string
+                              host_body_mass_index:
+                                anyof:
+                                  - min_exclusive: 0
+                                    type: number
+                                  - allowed:
+                                    - not applicable
+                                    - not collected
+                                    - not provided
+                                    - restricted access
+                                    type: string
+                                empty: false
+                                field_desc: Body mass index
+                                is_phi: false
+                                required: true
+                              host_common_name:
+                                allowed:
+                                  - human
+                                default: human
+                                type: string
+                              host_height:
+                                anyof:
+                                  - min_exclusive: 0
+                                    type: number
+                                  - allowed:
+                                    - not collected
+                                    - not provided
+                                    type: string
+                                empty: false
+                                field_desc: Height (or length) of host in cm preferred
+                                is_phi: false
+                                required: true
+                                units: host_height_units
+                              host_height_units:
+                                allowed:
+                                  - cm
+                                default: cm
+                                empty: false
+                                is_phi: false
+                                required: true
+                                type: string
+                              host_scientific_name:
+                                allowed:
+                                  - Homo sapiens
+                                default: Homo sapiens
+                                type: string
+                              host_taxid:
+                                allowed:
+                                  - '9606'
+                                default: '9606'
+                                type: string
+                              host_weight_units:
+                                allowed:
+                                  - kg
+                                default: kg
+                                type: string
+                              irb_institute:
+                                type: string
+                                empty: false
+                                field_desc: the name of institute that approved the IRB (human studies)
+                                is_phi: false
+                                required: false
+                              irb_protocol_id:
+                                type: string
+                                empty: false
+                                field_desc: the ID of the IRB protocol that covers the collection and analysis of
+                                  this sample (human studies)
+                                is_phi: false
+                                required: false
+                            sample_type_specific_metadata:
+                              antecubital crease:
+                                alias: "antecubital fossa"
+                              antecubital fossa:
+                                # TODO: this is a duplicate of "skin of elbow"
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:elbow
+                                    default: UBERON:elbow
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of arm"
+                                    default: "skin of arm"
+                                    type: string
+                              axilla skin:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:axilla
+                                    default: UBERON:axilla
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "axilla skin"
+                                    default: "axilla skin"
+                                    type: string
+                              back:
+                                alias: "skin of back"
+                              buccal mucosa:
+                                base_type: "saliva"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:buccal mucosa
+                                    default: UBERON:buccal mucosa
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "buccal mucosa"
+                                    default: "buccal mucosa"
+                                    type: string
+                              cheek:
+                                alias: "skin of cheek"
+                              # TODO: why is this a different item than (rather than an alias of) "skin of elbow"?
+                              elbow skin:
+                                alias: "skin of elbow"
+                              external auditory canal:
+                                alias: "skin of external ear canal"
+                              face:
+                                alias: "skin of face"
+                              feces:
+                                metadata_fields:
+                                  env_package:
+                                    allowed:
+                                      - human-gut
+                                    default: human-gut
+                                    type: string
+                                  scientific_name:
+                                    allowed:
+                                      - human gut metagenome
+                                    default: human gut metagenome
+                                    type: string
+                                  taxon_id:
+                                    allowed:
+                                      - 408170
+                                    default: 408170
+                                    type: integer
+                              forehead:
+                                alias: "skin of forehead"
+                              glabella:
+                                alias: "glabella skin"
+                              glabella skin:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:glabella skin
+                                    default: UBERON:glabella skin
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of forehead"
+                                    default: "skin of forehead"
+                                    type: string
+                              heel skin:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:heel
+                                    default: UBERON:heel
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of foot"
+                                    default: "skin of foot"
+                                    type: string
+                              hypothenar palm:
+                                alias: "skin of palm of manus"
+                              inguinal crease:
+                                alias: "skin of inguinal fold"
+                              interdigit web:
+                                alias: "manual digit skin"
+                              manual digit skin:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:manual digit
+                                    default: UBERON:manual digit
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of hand"
+                                    default: "skin of hand"
+                                    type: string
+                              manubrium:
+                                alias: "skin of manubrium of sternum"
+                              nail of pedal digit:
+                                base_type: "skin"
+                                metadata_fields:
+                                  env_material:
+                                    allowed:
+                                      - organic material
+                                    default: organic material
+                                    type: string
+                                  env_package:
+                                    allowed:
+                                      - human-associated
+                                    default: human-associated
+                                    type: string
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:pedal digit
+                                    default: UBERON:pedal digit
+                                    type: string
+                                  host_body_product:
+                                    allowed:
+                                      - UBERON:nail
+                                    default: UBERON:nail
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "nail of pedal digit"
+                                    default: "nail of pedal digit"
+                                    type: string
+                              nose:
+                                alias: "nose skin"
+                              nose skin:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - "UBERON:external nose"
+                                    default: "UBERON:external nose"
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of face"
+                                    default: "skin of face"
+                                    type: string
+                              occiput:
+                                alias: "skin of occiput"
+                              oral mucosa:
+                                alias: "buccal mucosa"
+                              palm:
+                                alias: "skin of palm of manus"
+                              plantar heel:
+                                alias: "heel skin"
+                              pedal digit skin:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:pedal digit skin
+                                    default: UBERON:pedal digit skin
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of foot"
+                                    default: "skin of foot"
+                                    type: string
+                              popliteal area:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:popliteal area
+                                    default: UBERON:popliteal area
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of knee"
+                                    default: "skin of knee"
+                                    type: string
+                              popliteal crease:
+                                alias: "popliteal area"
+                              popliteal fossa:
+                                alias: "popliteal area"
+                              retroauricular crease:
+                                alias: "retroauricular region"
+                              retroauricular region:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:retroauricular area
+                                    default: UBERON:retroauricular area
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of external ear"
+                                    default: "skin of external ear"
+                                    type: string
+                              saliva:
+                                metadata_fields:
+                                  env_package:
+                                    allowed:
+                                      - human-oral
+                                    default: human-oral
+                                    type: string
+                                  scientific_name:
+                                    allowed:
+                                      - human oral metagenome
+                                    default: human oral metagenome
+                                    type: string
+                                  taxon_id:
+                                    allowed:
+                                      - 447426
+                                    default: 447426
+                                    type: integer
+                              scalp:
+                                alias: "skin of scalp"
+                              skin:
+                                metadata_fields:
+                                  env_package:
+                                    allowed:
+                                      - human-skin
+                                    default: human-skin
+                                    type: string
+                                  scientific_name:
+                                    allowed:
+                                      - human skin metagenome
+                                    default: human skin metagenome
+                                    type: string
+                                  taxon_id:
+                                    allowed:
+                                      - 539655
+                                    default: 539655
+                                    type: integer
+                              skin of back:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - "UBERON:skin of back"
+                                    default: "UBERON:skin of back"
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of trunk"
+                                    default: "skin of trunk"
+                                    type: string
+                              skin of breast:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:breast
+                                    default: UBERON:breast
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin"
+                                    default: "skin"
+                                    type: string
+                              skin of cheek:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:cheek
+                                    default: UBERON:cheek
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of head"
+                                    default: "skin of head"
+                                    type: string
+                              skin of elbow:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:elbow
+                                    default: UBERON:elbow
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of arm"
+                                    default: "skin of arm"
+                                    type: string
+                              skin of external ear:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:ear
+                                    default: UBERON:ear
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of external ear"
+                                    default: "skin of external ear"
+                                    type: string
+                              skin of external ear canal:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:external acoustic meatus
+                                    default: UBERON:external acoustic meatus
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of external ear"
+                                    default: "skin of external ear"
+                                    type: string
+                              skin of face:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:face
+                                    default: UBERON:face
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of face"
+                                    default: "skin of face"
+                                    type: string
+                              skin of forearm:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:skin of forearm
+                                    default: UBERON:skin of forearm
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of arm"
+                                    default: "skin of arm"
+                                    type: string
+                              # TODO: for ABTX, the qiita_sample_type for skin of forehead is "skin",
+                              #  but for TRPCA it is given as "skin of forehead"; why?
+                              skin of forehead:
+                                base_type: "skin"
+                                metadata_fields:
+                                  # TODO: for ABTX, the host_body_site for skin of forehead is "UBERON:skin of forehead",
+                                  #  but for TRPCA it is given as "UBERON:forehead"; why?
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:skin of forehead
+                                    default: UBERON:skin of forehead
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin"
+                                    default: "skin"
+                                    type: string
+                              skin of hand:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:skin of hand
+                                    default: UBERON:skin of hand
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin"
+                                    default: "skin"
+                                    type: string
+                              skin of inguinal fold:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:inguinal fold
+                                    default: UBERON:inguinal fold
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "inguinal skin region"
+                                    default: "inguinal skin region"
+                                    type: string
+                              skin of manubrium of sternum:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:manubrium of sternum
+                                    default: UBERON:manubrium of sternum
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of chest"
+                                    default: "skin of chest"
+                                    type: string
+                              skin of occiput:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:surface of occiput
+                                    default: UBERON:surface of occiput
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of head"
+                                    default: "skin of head"
+                                    type: string
+                              skin of palm of hypothenar manus:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:hypothenar eminence
+                                    default: UBERON:hypothenar eminence
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of hand"
+                                    default: "skin of hand"
+                                    type: string
+                              skin of palm of manus :
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:palmar part of manus
+                                    default: UBERON:palmar part of manus
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of hand"
+                                    default: "skin of hand"
+                                    type: string
+                              skin of scalp:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:scalp
+                                    default: UBERON:scalp
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of head"
+                                    default: "skin of head"
+                                    type: string
+                              skin of umbilical area:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:umbilicus
+                                    default: UBERON:umbilicus
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of trunk"
+                                    default: "skin of trunk"
+                                    type: string
+                              thigh:
+                                alias: "upper leg skin"
+                              toe web space:
+                                alias: "pedal digit skin"
+                              toenail:
+                                alias: "nail of pedal digit"
+                              umbilicus:
+                                alias: "skin of umbilical area"
+                              upper leg skin:
+                                base_type: "skin"
+                                metadata_fields:
+                                  host_body_site:
+                                    allowed:
+                                      - UBERON:upper leg skin
+                                    default: UBERON:upper leg skin
+                                    type: string
+                                  qiita_sample_type:
+                                    allowed:
+                                      - "skin of leg"
+                                    default: "skin of leg"
+                                    type: string
+                              volar forearm:
+                                alias: "skin of forearm"
+                          mouse:
+                            metadata_fields:
+                              env_biome:
+                                allowed:
+                                  - urban biome
+                                default: urban biome
+                                type: string
+                              host_age_units:
+                                allowed:
+                                  - weeks
+                                default: weeks
+                                type: string
+                              host_common_name:
+                                allowed:
+                                  - mouse
+                                default: mouse
+                                type: string
+                              host_scientific_name:
+                                allowed:
+                                  - Mus musculus
+                                default: Mus musculus
+                                type: string
+                              host_taxid:
+                                allowed:
+                                  - '10090'
+                                default: '10090'
+                                type: string
+                              host_weight_units:
+                                allowed:
+                                  - g
+                                default: g
+                                type: string
+          free_living:
+            metadata_fields:
+              empo_1:
+                allowed:
+                  - Free-living
+                default: Free-living
+                type: string
+            host_type_specific_metadata:
+              free_living_non_saline:
+                metadata_fields:
+                  empo_2:
+                    allowed:
+                      - Free-living (non-saline)
+                    default: Free-living (non-saline)
+                    type: string
+                host_type_specific_metadata:
+                  aqueous_non_saline:
+                    metadata_fields:
+                      empo_3:
+                        allowed:
+                          - Aqueous (non-saline)
+                        default: Aqueous (non-saline)
+                        type: string
+                    host_type_specific_metadata:
+                      aerosol_non_saline:
+                        metadata_fields:
+                          empo_4:
+                            allowed:
+                              - Aerosol (non-saline)
+                            default: Aerosol (non-saline)
+                            type: string
+                      aqueous_surface_non_saline:
+                        metadata_fields:
+                          empo_4:
+                            allowed:
+                              - Surface (non-saline)
+                            default: Surface (non-saline)
+                            type: string
+                      water_non_saline:
+                        metadata_fields:
+                          empo_4:
+                            allowed:
+                              - Water (non-saline)
+                            default: Water (non-saline)
+                            type: string
+                          env_package:
+                            allowed:
+                              - water
+                            default: water
+                            type: string
+                        sample_type_specific_metadata:
+                          fresh water:
+                            metadata_fields:
+                              env_biome:
+                                allowed:
+                                  - freshwater biome
+                                default: freshwater biome
+                                type: string
+                              env_material:
+                                allowed:
+                                  - fresh water
+                                default: fresh water
+                                type: string
+                              scientific_name:
+                                allowed:
+                                  - freshwater metagenome
+                                default: freshwater metagenome
+                                type: string
+                              taxon_id:
+                                allowed:
+                                  - 449393
+                                default: 449393
+                                type: integer
+                            sample_type_specific_metadata:
+                              # is filtered water always urban?
+                              filtered water:
+                                metadata_fields:
+                                  env_biome:
+                                    allowed:
+                                      - urban biome
+                                    default: urban biome
+                                    type: string
+                          waste water:
+                            metadata_fields:
+                              env_biome:
+                                allowed:
+                                  - urban biome
+                                default: urban biome
+                                type: string
+                              env_material:
+                                allowed:
+                                - waste water
+                                - sewage
+                                type: string
+                              env_package:
+                                allowed:
+                                  - wastewater/sludge
+                                default: wastewater/sludge
+                                type: string
+                              scientific_name:
+                                allowed:
+                                  - wastewater metagenome
+                                default: wastewater metagenome
+                                type: string
+                              taxon_id:
+                                allowed:
+                                  - 527639
+                                default: 527639
+                                type: integer
+                  solid_non_saline:
+                    metadata_fields:
+                      empo_3:
+                        allowed:
+                          - Solid (non-saline)
+                        default: Solid (non-saline)
+                        type: string
+                    host_type_specific_metadata:
+                      sediment_non_saline:
+                        metadata_fields:
+                          empo_4:
+                            allowed:
+                              - Sediment (non-saline)
+                            default: Sediment (non-saline)
+                            type: string
+                      soil_non_saline:
+                        metadata_fields:
+                          empo_4:
+                            allowed:
+                              - Soil (non-saline)
+                            default: Soil (non-saline)
+                            type: string
+                      solid_surface_non_saline:
+                        metadata_fields:
+                          empo_4:
+                            allowed:
+                              - Surface (non-saline)
+                            default: Surface (non-saline)
+                            type: string
+                        host_type_specific_metadata:
+                          control surface:
+                            metadata_fields:
+                              env_biome:
+                                allowed:
+                                  - urban biome
+                                default: urban biome
+                                type: string
+                              env_feature:
+                                allowed:
+                                  - research facility
+                                default: research facility
+                                type: string
+                              env_material:
+                                allowed:
+                                  - anthropogenic environmental material
+                                default: anthropogenic environmental material
+                                type: string
+                              env_package:
+                                allowed:
+                                  - misc environment
+                                default: misc environment
+                                type: string
+                              scientific_name:
+                                allowed:
+                                  - indoor metagenome
+                                default: indoor metagenome
+                                type: string
+                              taxon_id:
+                                allowed:
+                                  - 1256227
+                                default: 1256227
+                                type: integer
+                            sample_type_specific_metadata:
+                              control surface:
+                                metadata_fields:
+                                  qiita_sample_type:
+                                    allowed:
+                                      - control surface
+                                    default: control surface
+                                    type: string
+                              control glove:
+                                metadata_fields:
+                                  qiita_sample_type:
+                                    allowed:
+                                      - control glove
+                                    default: control glove
+                                    type: string
+                              control swab:
+                                metadata_fields:
+                                  qiita_sample_type:
+                                    allowed:
+                                      - control swab
+                                    default: control swab
+                                    type: string
+          control:
+            default: "not applicable"
+            metadata_fields:
+              empo_1:
+                allowed:
+                  - Control
+                default: Control
+                type: string
+              env_feature:
+                allowed:
+                  - research facility
+                default: research facility
+                type: string
+              env_package:
+                allowed:
+                  - misc environment
+                default: misc environment
+                type: string
+              scientific_name:
+                allowed:
+                  - metagenome
+                default: metagenome
+                type: string
+              taxon_id:
+                allowed:
+                  - 256318
+                default: 256318
+                type: integer
+            host_type_specific_metadata:
+              # NB: control surfaces are actually under free-living!
+              sterile_water_blank:
+                metadata_fields:
+                  empo_2:
+                    allowed:
+                      - Negative
+                    default: Negative
+                    type: string
+                  empo_3:
+                    allowed:
+                      - Sterile water blank
+                    default: Sterile water blank
+                    type: string
+                  empo_4:
+                    allowed:
+                      - Sterile water blank
+                    default: Sterile water blank
+                    type: string
+                  env_biome:
+                    allowed:
+                      - urban biome
+                    default: urban biome
+                    type: string
+                  env_material:
+                    allowed:
+                      - sterile water
+                    default: sterile water
+                    type: string
+                sample_type_specific_metadata:
+                  control blank:
+                    metadata_fields:
+                      qiita_sample_type:
+                        allowed:
+                          - control blank
+                        default: control blank
+                        type: string
+                  control blank DNA extraction:
+                    base_type: "control blank"
+                  control blank library prep:
+                    base_type: "control blank"
+                  control solvent:
+                    base_type: "control blank"
+                  control shield:
+                    base_type: "control blank"
+              single strain:
+                metadata_fields:
+                  empo_2:
+                    allowed:
+                      - Positive
+                    default: Positive
+                    type: string
+                  empo_3:
+                    allowed:
+                      - Single strain
+                    default: Single strain
+                    type: string
+                  empo_4:
+                    allowed:
+                      - Single strain
+                    default: Single strain
+                    type: string
+                  env_material:
+                    allowed:
+                      - cell culture
+                    default: cell culture
+                    type: string
+                sample_type_specific_metadata:
+                  control positive:
+                    metadata_fields:
+                      qiita_sample_type:
+                        allowed:
+                          - control positive
+                        default: control positive
+                        type: string
+                  control positive DNA extraction:
+                    base_type: "control positive"
+                  control positive library prep:
+                    base_type: "control positive"
+                  control positive katharoseq:
+                    base_type: "control positive"
+                  # TODO: check this with Gail; isn't in metadata template
+                  control soap:
+                    metadata_fields:
+                      env_material:
+                        allowed:
+                          - anthropogenic environmental material
+                        default: anthropogenic environmental material
+                      scientific_name:
+                        allowed:
+                          - indoor metagenome
+                        default: indoor metagenome
+                        type: string
+                      taxon_id:
+                        allowed:
+                          - 1256227
+                        default: 1256227
+                        type: integer
+                  # TODO: check this with Gail; isn't in metadata template
+                  control shield:
+                    metadata_fields:
+                      qiita_sample_type:
+                        allowed:
+                          - "control shield"
+                        default: "control shield"
+                        type: string