forked from MIRRI/mirri_utils
340 lines
11 KiB
Python
340 lines
11 KiB
Python
from pathlib import Path
|
||
|
||
DATA_DIR = Path(__file__).parent / "data"
|
||
|
||
ACCESSION_NUMBER = "accession_number"
|
||
MIRRI_ACCESSION_NUMBER = 'mirri_accession_number'
|
||
RESTRICTION_ON_USE = "restriction_on_use"
|
||
NAGOYA_PROTOCOL = "nagoya_protocol"
|
||
ABS_RELATED_FILES = "abs_related_files"
|
||
MTA_FILES = "mta_file"
|
||
OTHER_CULTURE_NUMBERS = "other_culture_collection_numbers"
|
||
STRAIN_FROM_REGISTERED_COLLECTION = "strain_from_a_registered_collection"
|
||
RISK_GROUP = "risk_group"
|
||
DUAL_USE = "dual_use"
|
||
QUARANTINE = "quarantine"
|
||
ORGANISM_TYPE = "organism_type"
|
||
TAXON_NAME = "taxon_name"
|
||
TYPE = "type"
|
||
INFRASUBSPECIFIC_NAME = "infrasubspecific_names"
|
||
COMMENTS_ON_TAXONOMY = "comments_on_taxonomy"
|
||
STATUS = "status"
|
||
HISTORY_OF_DEPOSIT = "history_of_deposit"
|
||
DEPOSITOR = "depositor"
|
||
DATE_OF_DEPOSIT = "date_of_deposit"
|
||
COLLECTED_BY = "collected_by"
|
||
DATE_OF_COLLECTION = "date_of_collection"
|
||
ISOLATED_BY = "isolated_by"
|
||
DATE_OF_ISOLATION = "date_of_isolation"
|
||
DATE_OF_INCLUSION = "date_of_inclusion_on_catalog"
|
||
TESTED_TEMPERATURE_GROWTH_RANGE = "tested_temperature_growth_range"
|
||
RECOMMENDED_GROWTH_TEMP = "recommended_growth_temperature"
|
||
RECOMMENDED_GROWTH_MEDIUM = "recommended_media_for_growth"
|
||
FORM_OF_SUPPLY = "form_of_supply"
|
||
GEO_COORDS = "coordinates_of_geographic_origin"
|
||
ACCESSION_NAME = "other_denomination"
|
||
ALTITUDE = "altitude_of_geographic_origin"
|
||
GEOGRAPHIC_ORIGIN = "geographic_origin"
|
||
GMO = "gmo"
|
||
GMO_CONSTRUCTION_INFO = "gmo_construction_information"
|
||
MUTANT_INFORMATION = "mutant_information"
|
||
GENOTYPE = "genotype"
|
||
LITERATURE = "literature"
|
||
SEXUAL_STATE = "sexual_state"
|
||
PLOIDY = "ploidy"
|
||
INTERSPECIFIC_HYBRID = "interspecific_hybrid"
|
||
HYBRIDS = 'hybrids'
|
||
PLANT_PATHOGENICITY_CODE = "plant_pathogenicity_code"
|
||
PATHOGENICITY = "pathogenicity"
|
||
ENZYME_PRODUCTION = "enzyme_production"
|
||
PRODUCTION_OF_METABOLITES = "production_of_metabolites"
|
||
APPLICATIONS = "applications"
|
||
REMARKS = "remarks"
|
||
PLASMIDS = "plasmids"
|
||
PLASMIDS_COLLECTION_FIELDS = "plasmids_collections_fields"
|
||
SUBSTRATE_HOST_OF_ISOLATION = "substrate_host_of_isolation"
|
||
ISOLATION_HABITAT = "isolation_habitat"
|
||
ONTOBIOTOPE_ISOLATION_HABITAT = "ontobiotope_term_for_the_isolation_habitat"
|
||
LITERATURE_LINKED_TO_SEQ_GENOME = "literature_linked_to_the_sequence_genome"
|
||
AXENIC_CULTURE = "axenic_culture"
|
||
QPS ="qps"
|
||
SITE_LINK = "site_links"
|
||
|
||
# StrainId
|
||
STRAIN_ID = "id"
|
||
COLLECTION_CODE = "collection_code"
|
||
STRAIN_PUI = "strain_pui"
|
||
STRAIN_URL = "strain_url"
|
||
|
||
ID_SYNONYMS = 'id_synonyms'
|
||
# Taxonomy
|
||
GENUS = "genus"
|
||
SPECIES = "species"
|
||
|
||
# Location
|
||
COUNTRY = "countryOfOriginCode"
|
||
SITE = "site"
|
||
STATE = "state"
|
||
PROVINCE = "province"
|
||
MUNICIPALITY = "municipality"
|
||
ISLAND = "island"
|
||
OTHER = "other"
|
||
LATITUDE = "latitude"
|
||
LONGITUDE = "longitude"
|
||
ALTITUDE = "altitude"
|
||
GEOREF_METHOD = "georeferencingMethod"
|
||
COORDUNCERTAINTY = "coordUncertainty"
|
||
COORD_SPATIAL_REFERENCE = "coordenatesSpatialReference"
|
||
LOCATION = "location"
|
||
|
||
ALLOWED_COLLECTING_SITE_KEYS = [
|
||
COUNTRY,
|
||
STATE,
|
||
PROVINCE,
|
||
ISLAND,
|
||
MUNICIPALITY,
|
||
OTHER,
|
||
SITE,
|
||
LATITUDE,
|
||
LONGITUDE,
|
||
ALTITUDE,
|
||
GEOREF_METHOD,
|
||
COORDUNCERTAINTY,
|
||
COORD_SPATIAL_REFERENCE,
|
||
]
|
||
|
||
MIRRI_FIELDS = [
|
||
{"attribute": "id", "label": "accessionNumber"},
|
||
{"attribute": "mirri_accession_number", "label": "mirriAccessionNumber"},
|
||
{"attribute": "qps", "label": "qps"},
|
||
{"attribute": "axenic_culture", "label": "axenicCulture"},
|
||
{"attribute": "restriction_on_use", "label": "useRestrictions"},
|
||
{"attribute": "nagoya_protocol",
|
||
"label": "nagoyaConditions"},
|
||
{"attribute": ABS_RELATED_FILES, "label": "absFile"},
|
||
{"attribute": "mta_files", "label": "mtaFile"},
|
||
{"attribute": "other_numbers", "label": "otherCollectionNumbers"},
|
||
{"attribute": "is_from_registered_collection",
|
||
"label": "registeredCollection"},
|
||
{"attribute": "risk_group", "label": "riskGroup"},
|
||
{"attribute": "is_potentially_harmful", "label": "dualUse"},
|
||
{"attribute": "is_subject_to_quarantine", "label": "euQuarantine"},
|
||
{"attribute": "taxonomy.organism_type", "label": "organismType"},
|
||
{"attribute": "taxonomy.taxon_name", "label": "speciesName"},
|
||
{"attribute": "taxonomy.infrasubspecific_name",
|
||
"label": "infrasubspecificNames"},
|
||
{"attribute": "taxonomy.comments", "label": "taxonomyComments"},
|
||
{"attribute": "taxonomy.interspecific_hybrid",
|
||
"label": "hybrid"},
|
||
{"attribute": "status", "label": "status"},
|
||
{"attribute": "history", "label": "depositHistory", },
|
||
{"attribute": "deposit.who", "label": "depositor"},
|
||
{"attribute": "deposit.date", "label": "depositDate"},
|
||
{"attribute": "catalog_inclusion_date",
|
||
"label": "accessionDate"},
|
||
{"attribute": "collect.who", "label": "collector"},
|
||
{"attribute": "collect.date", "label": "collectionDate"},
|
||
{"attribute": "isolation.who", "label": "isolator"},
|
||
{"attribute": "isolation.date", "label": "isolationDate"},
|
||
{"attribute": "isolation.substrate_host_of_isolation",
|
||
"label": "substrate"},
|
||
{"attribute": "growth.tested_temp_range",
|
||
"label": "temperatureGrowthRange"},
|
||
{"attribute": "growth.recommended_temp",
|
||
"label": "recommendedTemperature"},
|
||
{"attribute": "growth.recommended_media",
|
||
"label": "recommendedMedium"},
|
||
{"attribute": "form_of_supply", "label": "supplyForms"},
|
||
{"attribute": "other_denominations", "label": "otherDenomination"},
|
||
{"attribute": "collect.location.coords",
|
||
"label": "geographicCoordinates"},
|
||
{"attribute": "collect.site.links",
|
||
"label": "siteLinks"},
|
||
{"attribute": "collect.location.altitude",
|
||
"label": "country"},
|
||
{"attribute": "collect.location", "label": "geographicOrigin"},
|
||
{"attribute": "collect.habitat", "label": "isolationHabitat"},
|
||
{"attribute": "collect.habitat_ontobiotope",
|
||
"label": "ontobiotopeTerms"},
|
||
{"attribute": "genetics.gmo", "label": "gmo"},
|
||
{"attribute": "genetics.gmo_construction",
|
||
"label": "gmoConstruction"},
|
||
{"attribute": "genetics.mutant_info", "label": "mutant"},
|
||
{"attribute": "genetics.genotype", "label": "genotype"},
|
||
{"attribute": "genetics.sexual_state", "label": "sexualState"},
|
||
{"attribute": "genetics.ploidy", "label": "ploidy"},
|
||
{"attribute": "genetics.plasmids", "label": "plasmids"},
|
||
{"attribute": "genetics.plasmids_in_collections",
|
||
"label": "plasmidCollections"},
|
||
{"attribute": "publications", "label": "identificationLiterature"},
|
||
{"attribute": PLANT_PATHOGENICITY_CODE, "label": "Plant pathogenicity code"},
|
||
{"attribute": "pathogenicity", "label": "pathogenicity"},
|
||
{"attribute": "enzyme_production", "label": "enzymes"},
|
||
{"attribute": "production_of_metabolites",
|
||
"label": "metabolites"},
|
||
{"attribute": "type",
|
||
"label": "type"},
|
||
{"attribute": "applications", "label": "applications", },
|
||
{"attribute": "remarks", "label": "remarks"},
|
||
{"attribute": LITERATURE_LINKED_TO_SEQ_GENOME,
|
||
"label": "sequenceLiterature"},
|
||
]
|
||
|
||
ALLOWED_SUBTAXA = ["subspecies", "variety", "convarietas", "group", "forma",
|
||
'forma.specialis']
|
||
ALLOWED_TAXONOMIC_RANKS = ["family", "genus", "species"] + ALLOWED_SUBTAXA
|
||
|
||
# nagoya
|
||
NAGOYA_NO_RESTRICTIONS = "no_known_restrictions_under_the_Nagoya_protocol"
|
||
NAGOYA_DOCS_AVAILABLE = "documents_providing_proof_of_legal_access_and_terms_of_use_available_at_the_collection"
|
||
NAGOYA_PROBABLY_SCOPE = "strain_probably_in_scope,_please_contact_the_culture_collection"
|
||
|
||
ALLOWED_NAGOYA_OPTIONS = [NAGOYA_NO_RESTRICTIONS,
|
||
NAGOYA_DOCS_AVAILABLE, NAGOYA_PROBABLY_SCOPE]
|
||
|
||
# Use restriction
|
||
NO_RESTRICTION = "no_restriction"
|
||
ONLY_RESEARCH = "only_research"
|
||
COMMERCIAL_USE_WITH_AGREEMENT = "commercial_use_with_agreement"
|
||
|
||
ALLOWED_RESTRICTION_USE_OPTIONS = [
|
||
NO_RESTRICTION,
|
||
ONLY_RESEARCH,
|
||
COMMERCIAL_USE_WITH_AGREEMENT,
|
||
]
|
||
|
||
ALLOWED_RISK_GROUPS = ["1", "2", "3", "4"]
|
||
|
||
AGAR = "Agar"
|
||
CRYO = "Cryo"
|
||
DRY_ICE = "Dry Ice"
|
||
LIQUID_CULTURE_MEDIUM = "Liquid Culture Medium"
|
||
LYO = "Lyo"
|
||
OIL = "Oil"
|
||
WATER = "Water"
|
||
ALLOWED_FORMS_OF_SUPPLY = [AGAR, CRYO, DRY_ICE,
|
||
LIQUID_CULTURE_MEDIUM, LYO, OIL, WATER]
|
||
|
||
DEPOSIT = "deposit"
|
||
ISOLATION = "isolation"
|
||
COLLECT = "collect"
|
||
GROWTH = "growth"
|
||
GENETICS = "genetics"
|
||
TAXONOMY = "taxonomy"
|
||
# Markers
|
||
MARKERS = "markers"
|
||
MARKER_TYPE = "marker_type"
|
||
MARKER_INSDC = "INSDC"
|
||
MARKER_SEQ = "marker_seq"
|
||
ALLOWED_MARKER_TYPES = [
|
||
{"acronym": "16S rRNA", "marker": "16S rRNA"},
|
||
{"acronym": "ACT", "marker": "Actin"},
|
||
{"acronym": "CaM", "marker": "Calmodulin"},
|
||
{"acronym": "EF-1α", "marker": "elongation factor 1-alpha (EF-1α)"},
|
||
{"acronym": "ITS",
|
||
"marker": "nuclear ribosomal Internal Transcribed Spacer (ITS)"},
|
||
{"acronym": "LSU", "marker": "nuclear ribosomal Large SubUnit (LSU)"},
|
||
{"acronym": "RPB1", "marker": "Ribosomal RNA-coding genes RPB1"},
|
||
{"acronym": "RPB2", "marker": "Ribosomal RNA-coding genes RPB2"},
|
||
{"acronym": "TUBB", "marker": "β-Tubulin"},
|
||
]
|
||
|
||
PUBLICATIONS = "publications"
|
||
PUB_ID = "pub_id"
|
||
PUB_DOI = "pub_doi"
|
||
PUB_PMID = "pub_pmid"
|
||
PUB_PUBMED_ID = ''
|
||
PUB_FULL_REFERENCE = "full_reference"
|
||
PUB_TITLE = "title"
|
||
PUB_AUTHORS = "authors"
|
||
PUB_JOURNAL = "journal"
|
||
PUB_YEAR = "year"
|
||
PUB_VOLUME = "volume"
|
||
PUB_ISSUE = "issue"
|
||
PUB_FIRST_PAGE = "first_page"
|
||
PUB_LAST_PAGE = "last_page"
|
||
BOOK_TITLE = "book_title"
|
||
BOOK_EDITOR = "book_editor"
|
||
BOOK_PUBLISHER = "book_publisher"
|
||
|
||
|
||
PUBLICATION_FIELDS = [
|
||
{"label": "ID", "attribute": PUB_ID},
|
||
{"label": "PMID", "attribute": PUB_PMID},
|
||
{"label": "DOI", "attribute": PUB_DOI},
|
||
{"label": "Full reference", "attribute": PUB_FULL_REFERENCE},
|
||
{"label": "Authors", "attribute": PUB_AUTHORS},
|
||
{"label": "Title", "attribute": PUB_TITLE},
|
||
{"label": "Journal", "attribute": PUB_JOURNAL},
|
||
{"label": "Year", "attribute": PUB_YEAR},
|
||
{"label": "Volume", "attribute": PUB_VOLUME},
|
||
{"label": "Issue", "attribute": PUB_ISSUE},
|
||
{"label": "First page", "attribute": PUB_FIRST_PAGE},
|
||
{"label": "Last page", "attribute": PUB_FIRST_PAGE},
|
||
{"label": "Book title", "attribute": BOOK_TITLE},
|
||
{"label": "Editors", "attribute": BOOK_EDITOR},
|
||
{"label": "Publisher", "attribute": BOOK_PUBLISHER},
|
||
]
|
||
|
||
|
||
# ploidy
|
||
ANEUPLOID = 0
|
||
HAPLOID = 1
|
||
DIPLOID = 2
|
||
TRIPLOID = 3
|
||
TETRAPLOID = 4
|
||
POLYPLOID = 9
|
||
|
||
ALLOWED_PLOIDIES = [ANEUPLOID, HAPLOID, DIPLOID, TRIPLOID, TETRAPLOID,
|
||
POLYPLOID]
|
||
|
||
SUBTAXAS = {
|
||
"subsp.": "subspecies",
|
||
"var.": "variety",
|
||
"convar.": "convarietas",
|
||
"group.": "group",
|
||
"f.": "forma",
|
||
"f.sp.": "forma.specialis"
|
||
}
|
||
|
||
#Control
|
||
VERSION = "Version"
|
||
DATE = "Date"
|
||
|
||
|
||
#Country codes
|
||
COUNTRY = "Country"
|
||
CODE = "Code"
|
||
ADDITIONAL_INFORMATION_ON_THE_COUNTRY_OR_CODE = "Additional information on the country or code"
|
||
|
||
|
||
#Country codes files
|
||
COUNTRY_CODES_SHEET = [
|
||
{"label": "Country", "attribute": COUNTRY},
|
||
{"label": "Code", "attribute": CODE},
|
||
{"label": "Additional information on the country or code", "attribute": ADDITIONAL_INFORMATION_ON_THE_COUNTRY_OR_CODE},
|
||
]
|
||
|
||
|
||
#Controle files
|
||
CONTROL_FIELDS = [
|
||
{"label": "Version", "attribute": VERSION},
|
||
{"label": "Date", "attribute": DATE},
|
||
]
|
||
|
||
# Excel sheet name
|
||
LOCATIONS = "Geographic origin" # 'Locations'
|
||
GROWTH_MEDIA = "Growth media"
|
||
GENOMIC_INFO = "Genomic information"
|
||
STRAINS = "Strains"
|
||
LITERATURE_SHEET = "Literature"
|
||
SEXUAL_STATE_SHEET = "Sexual state"
|
||
RESOURCE_TYPES_VALUES = "Resource types values"
|
||
FORM_OF_SUPPLY_SHEET = "Forms of supply"
|
||
PLOIDY_SHEET = "Ploidy"
|
||
ONTOBIOTOPE = "Ontobiotope"
|
||
MARKERS = "Markers"
|
||
CONTROL_SHEET = "Version"
|
||
COUNTRY_CODES_SHEET = "Country codes"
|
||
RESOURCE_SHEET = 'Resource types values'
|