diff options
| -rw-r--r-- | bh20sequploader/bh20seq-schema.yml | 12 | ||||
| -rw-r--r-- | example/metadata.yaml | 6 | ||||
| -rw-r--r-- | example/minimal_example.yaml | 2 | ||||
| -rw-r--r-- | scripts/dict_ontology_standardization/ncbi_countries.csv | 29 | ||||
| -rw-r--r-- | scripts/dict_ontology_standardization/ncbi_sequencing_technology.csv | 6 | ||||
| -rw-r--r-- | scripts/dict_ontology_standardization/ncbi_speciesman_source.csv | 3 |
6 files changed, 48 insertions, 10 deletions
diff --git a/bh20sequploader/bh20seq-schema.yml b/bh20sequploader/bh20seq-schema.yml index c48ebce..64008f2 100644 --- a/bh20sequploader/bh20seq-schema.yml +++ b/bh20sequploader/bh20seq-schema.yml @@ -78,24 +78,26 @@ $graph: type: string? jsonldPredicate: _id: http://purl.obolibrary.org/obo/OBI_0001479 + _type: "@id" specimen_source2: doc: Method how the specimen was derived as NCIT IRI, e.g. http://purl.obolibrary.org/obo/NCIT_C155835 (=throat swabb) type: string? jsonldPredicate: _id: http://purl.obolibrary.org/obo/OBI_0001479 + _type: "@id" collection_date: doc: Date when the sample was taken - type: string? + type: string jsonldPredicate: _id: http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#C25164 collection_location: doc: Geographical location where the sample was collected as wikidata reference, e.g. http://www.wikidata.org/entity/Q148 (China) - type: string? + type: string jsonldPredicate: _id: http://purl.obolibrary.org/obo/GAZ_00000448 _type: "@id" sample_storage_conditions: - doc: Information aboout storage of a specified type, e.g. frozen specimen, paraffin, fresh .... + doc: Information about storage of a specified type, e.g. frozen specimen, paraffin, fresh .... type: string? jsonldPredicate: _id: http://purl.obolibrary.org/obo/OBI_0001472 @@ -138,11 +140,13 @@ $graph: type: string jsonldPredicate: _id: http://purl.obolibrary.org/obo/OBI_0600047 + _type: "@id" sample_sequencing_technology2: doc: Technology that was used to sequence this sample (e.g Sanger, Nanopor MiniION) type: string? jsonldPredicate: _id: http://purl.obolibrary.org/obo/OBI_0600047 + _type: "@id" sequence_assembly_method: doc: Protocol which provides instructions on the alignment of sequencing reads to reference genome type: string? @@ -210,6 +214,8 @@ $graph: type: string? jsonldPredicate: _id: http://semanticscience.org/resource/SIO_000115 + _type: "@id" + noLinkCheck: true - name: MainSchema type: record diff --git a/example/metadata.yaml b/example/metadata.yaml index 5e1be0a..8bbf980 100644 --- a/example/metadata.yaml +++ b/example/metadata.yaml @@ -15,9 +15,9 @@ sample: sample_id: Id of the sample as defined by the submitter collector_name: Name of the person that took the sample collecting_institution: Institute that was responsible of sampeling - specimen_source: Method how specimen was derived - specimen_source2: Method how specimen was derived if two methods were used - collection_date: Date when the sample was taken + specimen_source: http://purl.obolibrary.org/obo/NCIT_C155831 + specimen_source2: http://purl.obolibrary.org/obo/NCIT_C155835 + collection_date: 2020-01-01 collection_location: http://www.wikidata.org/entity/Q148 sample_storage_conditions: XXX additional_collection_information: XXX diff --git a/example/minimal_example.yaml b/example/minimal_example.yaml index f4118d1..ed578e2 100644 --- a/example/minimal_example.yaml +++ b/example/minimal_example.yaml @@ -8,6 +8,8 @@ sample: sample_id: XX collector_name: John Doe collecting_institution: Doe university + collection_date: 2020-01 + collection_location: http://www.wikidata.org/entity/Q148 virus: virus_species: http://purl.obolibrary.org/obo/NCBITaxon_2697049 diff --git a/scripts/dict_ontology_standardization/ncbi_countries.csv b/scripts/dict_ontology_standardization/ncbi_countries.csv index a3a55ca..d5e2235 100644 --- a/scripts/dict_ontology_standardization/ncbi_countries.csv +++ b/scripts/dict_ontology_standardization/ncbi_countries.csv @@ -1,4 +1,6 @@ 30.59 N 114.3 E,http://www.wikidata.org/entity/Q11746 +35.92 N 74.33 E,http://www.wikidata.org/entity/Q609024 +39.54 N 116.23 E,http://www.wikidata.org/entity/Q198244 Afghanistan,http://www.wikidata.org/entity/Q889 Albania,http://www.wikidata.org/entity/Q222 Algeria,http://www.wikidata.org/entity/Q262 @@ -9,6 +11,7 @@ Argentina,http://www.wikidata.org/entity/Q414 Armenia,http://www.wikidata.org/entity/Q399 Australia,http://www.wikidata.org/entity/Q408 Australia: Queensland,http://www.wikidata.org/entity/Q36074 +Australia: Victoria,http://www.wikidata.org/entity/Q36687 Austria,http://www.wikidata.org/entity/Q40 Azerbaijan,http://www.wikidata.org/entity/Q227 Bahrain,http://www.wikidata.org/entity/Q398 @@ -41,13 +44,16 @@ China: Chongqing,http://www.wikidata.org/entity/Q11725 China: Fujian,http://www.wikidata.org/entity/Q41705 China: Gansu,http://www.wikidata.org/entity/Q42392 China: Guangdong,http://www.wikidata.org/entity/Q15175 +"China: Guangdong, Guangzhou",http://www.wikidata.org/entity/Q16572 China: Guangxi Zhuang Autonomous Region,http://www.wikidata.org/entity/Q15176 +China: Guangzhou,http://www.wikidata.org/entity/Q16572 China: Guizhou,http://www.wikidata.org/entity/Q47097 China: Hainan,http://www.wikidata.org/entity/Q42200 China: Hebei,http://www.wikidata.org/entity/Q21208 China: Heilongjiang,http://www.wikidata.org/entity/Q19206 China: Henan,http://www.wikidata.org/entity/Q43684 China: Hong Kong,http://www.wikidata.org/entity/Q8646 +China: HuaShang,http://www.wikidata.org/entity/Q148 China: Hubei,http://www.wikidata.org/entity/Q46862 "China: Hubei, Wuhan",http://www.wikidata.org/entity/Q11746 China: Hunan,http://www.wikidata.org/entity/Q45761 @@ -63,13 +69,18 @@ China: Shaanxi,http://www.wikidata.org/entity/Q47974 China: Shandong,http://www.wikidata.org/entity/Q43407 China: Shanghai,http://www.wikidata.org/entity/Q8686 China: Shanxi,http://www.wikidata.org/entity/Q46913 +China: Shenzhen,http://www.wikidata.org/entity/Q15174 China: Sichuan,http://www.wikidata.org/entity/Q19770 China: Tianjin,http://www.wikidata.org/entity/Q11736 China: Tibet Autonomous Region,http://www.wikidata.org/entity/Q17269 +China: Wuhan,http://www.wikidata.org/entity/Q11746 +China:Wuhan,http://www.wikidata.org/entity/Q11746 China: Xinjiang,http://www.wikidata.org/entity/Q34800 China: Yunnan,http://www.wikidata.org/entity/Q43194 China: Zhejiang,http://www.wikidata.org/entity/Q16967 +"China: Zhejiang, Hangzhou",http://www.wikidata.org/entity/Q4970 Colombia,http://www.wikidata.org/entity/Q739 +Colombia: Antioquia,http://www.wikidata.org/entity/Q123304 Comoros,http://www.wikidata.org/entity/Q970 Costa Rica,http://www.wikidata.org/entity/Q800 Croatia,http://www.wikidata.org/entity/Q224 @@ -110,8 +121,10 @@ Hungary,http://www.wikidata.org/entity/Q28 Iceland,http://www.wikidata.org/entity/Q189 Icelandic Commonwealth,http://www.wikidata.org/entity/Q62389 India,http://www.wikidata.org/entity/Q668 +India: Kerala State,http://www.wikidata.org/entity/Q1186 Indonesia,http://www.wikidata.org/entity/Q252 Iran,http://www.wikidata.org/entity/Q794 +Iran: Qum,http://www.wikidata.org/entity/Q131664 Iran: Tehran,http://www.wikidata.org/entity/Q3616 Iraq,http://www.wikidata.org/entity/Q796 Ireland,http://www.wikidata.org/entity/Q27 @@ -163,12 +176,15 @@ New Zealand,http://www.wikidata.org/entity/Q664 Nicaragua,http://www.wikidata.org/entity/Q811 Niger,http://www.wikidata.org/entity/Q1032 Nigeria,http://www.wikidata.org/entity/Q1033 +Nigeria: Lagos,http://www.wikidata.org/entity/Q8673 North Korea,http://www.wikidata.org/entity/Q423 North Macedonia,http://www.wikidata.org/entity/Q221 Norway,http://www.wikidata.org/entity/Q20 Oman,http://www.wikidata.org/entity/Q842 Ottoman Empire,http://www.wikidata.org/entity/Q12560 Pakistan,http://www.wikidata.org/entity/Q843 +Pakistan: Gilgit,http://www.wikidata.org/entity/Q609024 +Pakistan: KPK,http://www.wikidata.org/entity/Q183314 Palau,http://www.wikidata.org/entity/Q695 Panama,http://www.wikidata.org/entity/Q804 Papua New Guinea,http://www.wikidata.org/entity/Q691 @@ -203,6 +219,7 @@ Slovenia,http://www.wikidata.org/entity/Q215 Solomon Islands,http://www.wikidata.org/entity/Q685 Somalia,http://www.wikidata.org/entity/Q1045 South Africa,http://www.wikidata.org/entity/Q258 +South Africa: KwaZulu-Natal,http://www.wikidata.org/entity/Q81725 South African Republic,http://www.wikidata.org/entity/Q550374 South Korea,http://www.wikidata.org/entity/Q884 South Sudan,http://www.wikidata.org/entity/Q958 @@ -215,6 +232,7 @@ Suriname,http://www.wikidata.org/entity/Q730 Sweden,http://www.wikidata.org/entity/Q34 Switzerland,http://www.wikidata.org/entity/Q39 Syria,http://www.wikidata.org/entity/Q858 +Taiwan,http://www.wikidata.org/entity/Q865 Tajikistan,http://www.wikidata.org/entity/Q863 Tanzania,http://www.wikidata.org/entity/Q924 Thailand,http://www.wikidata.org/entity/Q869 @@ -224,6 +242,7 @@ Togo,http://www.wikidata.org/entity/Q945 Tonga,http://www.wikidata.org/entity/Q678 Trinidad and Tobago,http://www.wikidata.org/entity/Q754 Tunisia,http://www.wikidata.org/entity/Q948 +Tunisia: Tunis,http://www.wikidata.org/entity/Q3572 Turkey,http://www.wikidata.org/entity/Q43 Turkmenistan,http://www.wikidata.org/entity/Q874 Tuvalu,http://www.wikidata.org/entity/Q672 @@ -240,6 +259,7 @@ USA: AL,http://www.wikidata.org/entity/Q173 USA: AR,http://www.wikidata.org/entity/Q1612 USA: AZ,http://www.wikidata.org/entity/Q816 USA: CA,http://www.wikidata.org/entity/Q99 +"USA: CA, San Diego County",http://www.wikidata.org/entity/Q108143 USA: CO,http://www.wikidata.org/entity/Q1261 USA: CT,http://www.wikidata.org/entity/Q779 USA: DE,http://www.wikidata.org/entity/Q1393 @@ -249,6 +269,7 @@ USA: HI,http://www.wikidata.org/entity/Q782 USA: IA,http://www.wikidata.org/entity/Q1546 USA: ID,http://www.wikidata.org/entity/Q1221 USA: IL,http://www.wikidata.org/entity/Q1204 +USA: Illinois,http://www.wikidata.org/entity/Q1204 USA: IN,http://www.wikidata.org/entity/Q1415 USA: KS,http://www.wikidata.org/entity/Q1558 USA: KY,http://www.wikidata.org/entity/Q1603 @@ -267,6 +288,7 @@ USA: NE,http://www.wikidata.org/entity/Q1553 USA: NH,http://www.wikidata.org/entity/Q759 USA: NJ,http://www.wikidata.org/entity/Q1408 USA: NM,http://www.wikidata.org/entity/Q1522 +USA: North Carolina,http://www.wikidata.org/entity/Q1454 USA: NV,http://www.wikidata.org/entity/Q1227 USA: NY,http://www.wikidata.org/entity/Q1384 USA: OH,http://www.wikidata.org/entity/Q1397 @@ -274,8 +296,10 @@ USA: OK,http://www.wikidata.org/entity/Q1649 USA: OR,http://www.wikidata.org/entity/Q824 USA: PA,http://www.wikidata.org/entity/Q1400 USA: RI,http://www.wikidata.org/entity/Q1387 +"USA: San Francisco, CA",http://www.wikidata.org/entity/Q62 USA: SC,http://www.wikidata.org/entity/Q1456 USA: SD,http://www.wikidata.org/entity/Q1211 +"USA: Snohomish County, WA",http://www.wikidata.org/entity/Q110403 USA: TN,http://www.wikidata.org/entity/Q1509 USA: TX,http://www.wikidata.org/entity/Q1439 USA: UT,http://www.wikidata.org/entity/Q829 @@ -289,9 +313,10 @@ Uzbekistan,http://www.wikidata.org/entity/Q265 Vanuatu,http://www.wikidata.org/entity/Q686 Vatican City,http://www.wikidata.org/entity/Q237 Venezuela,http://www.wikidata.org/entity/Q717 -Vietnam,http://www.wikidata.org/entity/Q881 Viet nam,http://www.wikidata.org/entity/Q881 +Viet Nam,http://www.wikidata.org/entity/Q881 Viet Nam: Ho Chi Minh city,http://www.wikidata.org/entity/Q1854 +Vietnam,http://www.wikidata.org/entity/Q881 Yemen,http://www.wikidata.org/entity/Q805 Zambia,http://www.wikidata.org/entity/Q953 -Zimbabwe,http://www.wikidata.org/entity/Q954
\ No newline at end of file +Zimbabwe,http://www.wikidata.org/entity/Q954 diff --git a/scripts/dict_ontology_standardization/ncbi_sequencing_technology.csv b/scripts/dict_ontology_standardization/ncbi_sequencing_technology.csv index 12d4e94..8e37b8c 100644 --- a/scripts/dict_ontology_standardization/ncbi_sequencing_technology.csv +++ b/scripts/dict_ontology_standardization/ncbi_sequencing_technology.csv @@ -5,6 +5,7 @@ Oxford Nanopore MinION,http://www.ebi.ac.uk/efo/EFO_0008632 ONT (Oxford Nanopore Technologies),http://www.ebi.ac.uk/efo/EFO_0008632 Oxford Nanopore technologies MinION,http://www.ebi.ac.uk/efo/EFO_0008632 MinION Oxford Nanopore,http://www.ebi.ac.uk/efo/EFO_0008632 +Nanopore,http://purl.obolibrary.org/obo/NCIT_C146818 Illumina MiSeq,http://www.ebi.ac.uk/efo/EFO_0004205 Illumina,http://purl.obolibrary.org/obo/OBI_0000759 Oxford Nanopore technology,http://purl.obolibrary.org/obo/NCIT_C146818 @@ -12,6 +13,7 @@ Oxford Nanopore Technologies,http://purl.obolibrary.org/obo/NCIT_C146818 Oxford Nanopore,http://purl.obolibrary.org/obo/NCIT_C146818 IonTorrent,http://purl.obolibrary.org/obo/NCIT_C125894 Ion Torrent X5Plus,http://purl.obolibrary.org/obo/NCIT_C125894 +ThermoFisher S5Plus,http://purl.obolibrary.org/obo/NCIT_C125894 Sanger dideoxy sequencing,http://purl.obolibrary.org/obo/NCIT_C19641 -MGISEQ 2000, http://virtual-bh/MGISEQ2000 -MGISEQ2000, http://virtual-bh/MGISEQ2000
\ No newline at end of file +MGISEQ 2000,http://virtual-bh/MGISEQ2000 +MGISEQ2000,http://virtual-bh/MGISEQ2000
\ No newline at end of file diff --git a/scripts/dict_ontology_standardization/ncbi_speciesman_source.csv b/scripts/dict_ontology_standardization/ncbi_speciesman_source.csv index fcd6c94..2905588 100644 --- a/scripts/dict_ontology_standardization/ncbi_speciesman_source.csv +++ b/scripts/dict_ontology_standardization/ncbi_speciesman_source.csv @@ -20,3 +20,6 @@ swab,http://purl.obolibrary.org/obo/NCIT_C13195 oral swab,http://purl.obolibrary.org/obo/NCIT_C13195 bronchoalveolar lavage,http://purl.obolibrary.org/obo/NCIT_C13195 sputum,http://purl.obolibrary.org/obo/NCIT_C13278 +aspirate,http://purl.obolibrary.org/obo/NCIT_C13347 +stool,http://purl.obolibrary.org/obo/NCIT_C13234 +serum,http://purl.obolibrary.org/obo/NCIT_C13325
\ No newline at end of file |