Merge pull request #90 from AndreaGuarracino/patch-21

genbank and sra scripts update, new terms in the ontology dictionaries

3 files changed, 59 insertions, 2 deletions

diff --git a/scripts/dict_ontology_standardization/ncbi_countries.csv b/scripts/dict_ontology_standardization/ncbi_countries.csv
index 204f7f2..4bea3ec 100644
--- a/scripts/dict_ontology_standardization/ncbi_countries.csv
+++ b/scripts/dict_ontology_standardization/ncbi_countries.csv
@@ -284,6 +284,7 @@ USA:CA,http://www.wikidata.org/entity/Q99
 USA: California,http://www.wikidata.org/entity/Q99
 USA:California,http://www.wikidata.org/entity/Q99
 "USA: CA, San Diego County",http://www.wikidata.org/entity/Q108143
+"USA: California, Monterey County",http://www.wikidata.org/entity/Q108072
 USA: CO,http://www.wikidata.org/entity/Q1261
 USA: CT,http://www.wikidata.org/entity/Q779
 USA: Connecticut,http://www.wikidata.org/entity/Q779
@@ -301,8 +302,9 @@ USA: IN,http://www.wikidata.org/entity/Q1415
 USA: KS,http://www.wikidata.org/entity/Q1558
 USA: KY,http://www.wikidata.org/entity/Q1603
 USA: LA,http://www.wikidata.org/entity/Q1588
-USA:Los Angeles,http://www.wikidata.org/entity/Q65
-"USA: New Orleans, LA",http://www.wikidata.org/entity/Q34404
+"USA: SLIDELL, LA, LA",https://www.wikidata.org/wiki/Q988156
+"USA: Slidell, LA, LA",https://www.wikidata.org/wiki/Q988156
+"USA: New Orleans, LA",https://www.wikidata.org/wiki/Q34404
 USA: MA,http://www.wikidata.org/entity/Q771
 USA: Massachusetts,http://www.wikidata.org/entity/Q771
 USA: MD,http://www.wikidata.org/entity/Q1391
@@ -336,6 +338,33 @@ USA: SC,http://www.wikidata.org/entity/Q1456
 USA: South Carolina,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: Washington,Snohomish County",http://www.wikidata.org/entity/Q110403
+"USA: Washington, Snohomish County",http://www.wikidata.org/entity/Q110403
+"USA: Washington,Skagit County",http://www.wikidata.org/entity/Q113892
+"USA: Washington, Skagit County",http://www.wikidata.org/entity/Q113892
+"USA: Washington,Pierce County",http://www.wikidata.org/entity/Q156459
+"USA: Washington, Pierce County",http://www.wikidata.org/entity/Q156459
+"USA: Washington,Mason County",http://www.wikidata.org/entity/Q111904
+"USA: Washington, Mason County",http://www.wikidata.org/entity/Q111904
+"USA: Washington,Kittitas County",http://www.wikidata.org/entity/Q111540
+"USA: Washington,King County",http://www.wikidata.org/entity/Q108861
+"USA: Washington, King County",http://www.wikidata.org/entity/Q108861
+"USA: King County,WA",http://www.wikidata.org/entity/Q108861
+"USA: Washington,Jefferson County",http://www.wikidata.org/entity/Q384737
+"USA: Washington,Grant County",http://www.wikidata.org/entity/Q281681
+"USA: Washington, Grant County",http://www.wikidata.org/entity/Q281681
+"USA: Washington,Franklin County",http://www.wikidata.org/entity/Q118716
+"USA: Washington, Franklin County",http://www.wikidata.org/entity/Q118716
+"USA: Washington,Clark County",http://www.wikidata.org/entity/Q156287
+"USA: Washington,Benton County",http://www.wikidata.org/entity/Q156216
+"USA: Washington, Benton County",http://www.wikidata.org/entity/Q156216
+"USA: Washington,Asotin County",http://www.wikidata.org/entity/Q156295
+"USA: Washington, Asotin County",http://www.wikidata.org/entity/Q156295
+"USA: Washington,Adams County",http://www.wikidata.org/entity/Q156273
+"USA: Washington, Adams County",http://www.wikidata.org/entity/Q156273
+"USA: Washington, Spokane County",http://www.wikidata.org/entity/Q485276
+"USA: Washington, Douglas County",http://www.wikidata.org/entity/Q156220
+"USA: Washington, Cowlitz County",http://www.wikidata.org/entity/Q156276
 USA: TN,http://www.wikidata.org/entity/Q1509
 USA: TX,http://www.wikidata.org/entity/Q1439
 USA: UT,http://www.wikidata.org/entity/Q829
@@ -347,6 +376,10 @@ USA: WA,http://www.wikidata.org/entity/Q1223
 USA: Washington,http://www.wikidata.org/entity/Q1223
 USA: WI,http://www.wikidata.org/entity/Q1537
 USA: Wisconsin,http://www.wikidata.org/entity/Q1537
+"USA: Washington,Yakima County",http://www.wikidata.org/entity/Q156629
+"USA: Washington, Yakima County",http://www.wikidata.org/entity/Q156629
+"USA: Washington,Whatcom County",http://www.wikidata.org/entity/Q156623
+"USA: Dane County, Wisconsin",http://www.wikidata.org/entity/Q502200
 USA: WV,http://www.wikidata.org/entity/Q1371
 USA: WY,http://www.wikidata.org/entity/Q1214
 Uzbekistan,http://www.wikidata.org/entity/Q265
diff --git a/scripts/dict_ontology_standardization/ncbi_host_species.csv b/scripts/dict_ontology_standardization/ncbi_host_species.csv
index bc6ac04..40572a3 100644
--- a/scripts/dict_ontology_standardization/ncbi_host_species.csv
+++ b/scripts/dict_ontology_standardization/ncbi_host_species.csv
@@ -5,5 +5,6 @@ sapiens,http://purl.obolibrary.org/obo/NCBITaxon_9606
 Mustela lutreola,http://purl.obolibrary.org/obo/NCBITaxon_9666
 Manis javanica,http://purl.obolibrary.org/obo/NCBITaxon_9974
 Felis catus,http://purl.obolibrary.org/obo/NCBITaxon_9685
+Felis catus; Domestic Shorthair,http://purl.obolibrary.org/obo/NCBITaxon_9685
 Panthera tigris jacksoni,http://purl.obolibrary.org/obo/NCBITaxon_419130
 Canis lupus familiaris,http://purl.obolibrary.org/obo/NCBITaxon_9615
diff --git a/scripts/dict_ontology_standardization/ncbi_sequencing_technology.csv b/scripts/dict_ontology_standardization/ncbi_sequencing_technology.csv
index 964cbf3..59c4800 100644
--- a/scripts/dict_ontology_standardization/ncbi_sequencing_technology.csv
+++ b/scripts/dict_ontology_standardization/ncbi_sequencing_technology.csv
@@ -13,12 +13,14 @@ Illumina NextSeq 550,http://www.ebi.ac.uk/efo/EFO_0008566
 NextSeq550,http://www.ebi.ac.uk/efo/EFO_0008566
 NextSeq 550,http://www.ebi.ac.uk/efo/EFO_0008566
 Illumina MiniSeq,http://www.ebi.ac.uk/efo/EFO_0008636
+Illumina NovaSeq,http://www.ebi.ac.uk/efo/EFO_0008637
 Illumina NovaSeq 6000,http://www.ebi.ac.uk/efo/EFO_0008637
 Nanopore MinION,http://www.ebi.ac.uk/efo/EFO_0008632
 Oxford Nanopore MinION,http://www.ebi.ac.uk/efo/EFO_0008632
 ONT (Oxford Nanopore Technologies),http://purl.obolibrary.org/obo/NCIT_C146818
 Oxford Nanopore Technology,http://purl.obolibrary.org/obo/NCIT_C146818
 Oxford Nanopore technologies MinION,http://www.ebi.ac.uk/efo/EFO_0008632
+Oxford Nanopore Sequencing,http://purl.obolibrary.org/obo/NCIT_C146818
 MinION Oxford Nanopore,http://www.ebi.ac.uk/efo/EFO_0008632
 MinION,http://www.ebi.ac.uk/efo/EFO_0008632
 Nanopore,http://purl.obolibrary.org/obo/NCIT_C146818
@@ -33,3 +35,24 @@ 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
+Illumina HiSeq X,http://www.ebi.ac.uk/efo/EFO_0008567
+ONT GridION X5,http://www.ebi.ac.uk/efo/EFO_0008633
+ONT PremethION,http://www.ebi.ac.uk/efo/EFO_0008634
+PacBio RS II,http://www.ebi.ac.uk/efo/EFO_0008631
+PacBio Sequel System,http://www.ebi.ac.uk/efo/EFO_0008630
+Illumina Genome Analyzer,http://www.ebi.ac.uk/efo/EFO_0004200
+Illumina Genome Analyzer II,http://www.ebi.ac.uk/efo/EFO_0004201
+Illumina Genome Analyzer IIx,http://www.ebi.ac.uk/efo/EFO_0004202
+454 GS 20 sequencer,http://www.ebi.ac.uk/efo/EFO_0004206
+454 GS FLX Titanium sequencer,http://www.ebi.ac.uk/efo/EFO_0004433
+454 GS FLX sequencer,http://www.ebi.ac.uk/efo/EFO_0004432
+454 GS Junior sequencer,http://www.ebi.ac.uk/efo/EFO_0004434
+454 GS sequencer,http://www.ebi.ac.uk/efo/EFO_0004431
+AB SOLiD 4 System,http://www.ebi.ac.uk/efo/EFO_0004438
+AB SOLiD 4hq System,http://www.ebi.ac.uk/efo/EFO_0004441
+AB SOLiD 5500,http://www.ebi.ac.uk/efo/EFO_0004440
+AB SOLiD 5500xl,http://www.ebi.ac.uk/efo/EFO_0004436
+AB SOLiD PI System,http://www.ebi.ac.uk/efo/EFO_0004437
+AB SOLiD System,http://www.ebi.ac.uk/efo/EFO_0004435
+AB SOLiD System 2.0,http://www.ebi.ac.uk/efo/EFO_0004442
+AB SOLiD System 3.0,http://www.ebi.ac.uk/efo/EFO_0004439