Refactor internal data management

data-raw/internal-DO_gs.R → data-raw/internal/DO_gs.R

@@ -1,5 +1,9 @@
-## code to save DO Google Sheet information internally (`.DO_gs`)
-devtools::load_all()
+## code to prepare `.DO_gs` internal dataset ##
+#
+# Serves as a reference for DO-related Google Sheets and relevant sheets (tabs)
+# for data retrieval
+
+rlang::check_installed("here")
 
 .DO_gs <- list(
   users = list(
@@ -12,4 +16,8 @@ devtools::load_all()
   )
 )
 
-use_data_internal(.DO_gs, overwrite = TRUE)
+saveRDS(
+  .DO_gs,
+  file = here::here("data-raw", "internal", "DO_gs.rds"),
+  compress = "bzip2"
+)

data-raw/internal/curation_opts.R

@@ -0,0 +1,37 @@
+## code to prepare `.curation_opts` internal dataset ##
+#
+# This dataset is updated from a a Google Sheet and serves as a schema for
+# establishment of curation templates and their conversion to robot templates
+
+rlang::check_installed(
+    c("dplyr", "googlesheets4", "here", "vroom")
+)
+
+
+out_dir <- here::here("data-raw", "internal")
+
+# save full schema for developer reference
+curation_opts <- googlesheets4::read_sheet(
+    "https://docs.google.com/spreadsheets/d/1Zn6p5xkVHUwbWe1N8FUa3fNcEkAOoE9P4ADb12f69hQ/edit",
+    sheet = "template_options",
+    col_types = "c"
+) |>
+    dplyr::filter(!is.na(.data$template))
+
+vroom::vroom_write(
+    curation_opts,
+    file.path(out_dir, "curation_opts.tsv"),
+    na = ""
+)
+
+# save internal data
+.curation_opts <- dplyr::select(
+    curation_opts,
+    dplyr::all_of(c("data_type", "template", "inclusion"))
+)
+
+saveRDS(
+    .curation_opts,
+    file = file.path(out_dir, "curation_opts.rds"),
+    compress = "bzip2"
+)

data-raw/internal/curation_opts.tsv

@@ -0,0 +1,66 @@
+data_type	template	inclusion	example	notes	optional_values	alternate_title	alternate_format	sparql	export_header	sparql_dt_motif
+id	ID	required manual	DOID:0080943	replaced older 'iri/curie' header for simplicity; id now covered by obo id field	IRI or CURIE	CI	IRI or CURIE; CI means Class IRI --> type will be CLASS_TYPE	?iri a owl:Class .	ID	
+iri/curie	ID	deprecated	DOID:0080943	replaced by 'id' header	IRI or CURIE			?iri a owl:Class .	IRI	
+label	A rdfs:label	required manual	46,XX sex reversal 5					?iri rdfs:label ?label .	LABEL	rdfs:label
+parent id	SC % SPLIT=|	required manual	DOID:0111760	"accepts CURIE or IRI; intended for only asserted subclass relationships between named classes
+--> separated from subclass anon for practical purposes - ROBOT template is the same, ROBOT export differs"	disease by infectious agent			"?iri rdfs:subClassOf ?parent .
+FILTER(!isBlank(?parent))"	SubClass Of [NAMED ID]	rdfs:subClassOf
+definition	A IAO:0000115	required manual	A 46,XX sex reversal that is characterized by genital virilization in 46,XX individuals, associated with congenital heart disease and variable somatic anomalies including blepharophimosis-ptosis-epicanthus inversus syndrome and congenital diaphragmatic hernia and that has_material_basis_in heterozygous mutation in the NR2F2 gene on chromosome 15q26.					?iri obo:IAO_0000115 ?definition .	obo:IAO_0000115	IAO:0000115
+definition source(s)	>A oboInOwl:hasDbXref SPLIT=|	required manual	url:https://pubmed.ncbi.nlm.nih.gov/29478779/					"!<<.definition_axiom>>!
+   oboInOwl:hasDbXref ?def_src ."		IAO:0000115-oboInOwl:hasDbXref
+definition source type(s)	>AI dc11:type SPLIT=|	optional manual	curator inference from journal publication	do not quote!!!	ECO codes, e.g. ECO:0007645			"!<<.definition_axiom>>!
+   dc:type ?src_type ."		IAO:0000115-dc:type
+xref(s)	A oboInOwl:hasDbXref SPLIT=|	optional manual	OMIM:618901					?iri oboInOwl:hasDbXref ?xref .	oboInOwl:hasDbXref	oboInOwl:hasDbXref
+skos mapping(s): exact	A skos:exactMatch SPLIT=|	optional manual	OMIM:618901	adds skos mappings as strings; current INCORRECT DO format			"should use IRIs and be as follows:
+AI skos:exactMatch SPLIT=|
+ - example input: https://omim.org/MIM:618901"	?iri skos:exactMatch ?skos_exact .	skos:exactMatch	skos:exactMatch
+skos mapping(s): close	A skos:closeMatch SPLIT=|	optional manual	OMIM:618901	adds skos mappings as strings; current INCORRECT DO format			"should use IRIs and be as follows:
+AI skos:closeMatch SPLIT=|
+ - example input: https://omim.org/MIM:618901"	?iri skos:closeMatch ?skos_close .	skos:closeMatch	skos:closeMatch
+skos mapping(s): broad	A skos:broadMatch SPLIT=|	optional manual	OMIM:PS613135	adds skos mappings as strings; current INCORRECT DO format			"should use IRIs and be as follows:
+AI skos:exactMatch SPLIT=|
+ - example input: https://omim.org/MIM:618901"	?iri skos:broadMatch ?skos_broad .	skos:broadMatch	skos:broadMatch
+skos mapping(s): narrow	A skos:narrowMatch SPLIT=|	optional manual	OMIM:618901	adds skos mappings as strings; current INCORRECT DO format			"should use IRIs and be as follows:
+AI skos:exactMatch SPLIT=|
+ - example input: https://omim.org/MIM:618901"	?iri skos:narrowMatch ?skos_narrow .	skos:narrowMatch	skos:narrowMatch
+skos mapping(s): related	A skos:relatedMatch SPLIT=|	optional manual		adds skos mappings as strings; current INCORRECT DO format			"should use IRIs and be as follows:
+AI skos:exactMatch SPLIT=|
+ - example input: https://omim.org/MIM:618901"	?iri skos:relatedMatch ?skos_related .	skos:relatedMatch	skos:relatedMatch
+synonym(s): exact	A oboInOwl:hasExactSynonym SPLIT=|	optional manual	hemangiosarcoma	do not quote!!!				"glueV: .synonym_stmt, syn_scope = ""Exact"""	oboInOwl:hasExactSynonym	oboInOwl:hasExactSynonym
+synonym(s): broad	A oboInOwl:hasBroadSynonym SPLIT=|	optional manual						"glueV: .synonym_stmt, syn_scope = ""Broad"""	oboInOwl:hasBroadSynonym	oboInOwl:hasBroadSynonym
+synonym(s): narrow	A oboInOwl:hasNarrowSynonym SPLIT=|	optional manual						"glueV: .synonym_stmt, syn_scope = ""Narrow"""	oboInOwl:hasNarrowSynonym	oboInOwl:hasNarrowSynonym
+synonym(s): related	A oboInOwl:hasRelatedSynonym SPLIT=|	optional manual						"glueV: .synonym_stmt, syn_scope = ""Related"""	oboInOwl:hasRelatedSynonym	oboInOwl:hasRelatedSynonym
+acronym(s): exact	A oboInOwl:hasExactSynonym SPLIT=|	optional manual	CAMRQ	"must be accompanied by ""acronym annotation"" header/template in the adjacent column to the rigth in robot template"				"glueV: .acronym_stmt, acronym_scope = ""Exact"""		oboInOwl:hasExactSynonym-OMO:0003012
+acronym(s): broad	A oboInOwl:hasBroadSynonym SPLIT=|	optional manual						"glueV: .acronym_stmt, acronym_scope = ""Broad"""		oboInOwl:hasBroadSynonym-OMO:0003012
+acronym(s): narrow	A oboInOwl:hasNarrowSynonym SPLIT=|	optional manual						"glueV: .acronym_stmt, acronym_scope = ""Narrow"""		oboInOwl:hasNarrowSynonym-OMO:0003012
+acronym(s): related	A oboInOwl:hasRelatedSynonym SPLIT=|	optional manual	DES	"must be accompanied by ""acronym annotation"" header/template in the adjacent column to the rigth in robot template"				"glueV: .acronym_stmt, acronym_scope = ""Related"""		oboInOwl:hasRelatedSynonym-OMO:0003012
+acronym annotation	>AI oboInOwl:hasSynonymType	optional auto	acronym							
+eq class	EC % SPLIT=|	optional manual	disease and ('has material basis in' some (Viruses or Bacteria or Eukaryota))					"?iri owl:equivalentClass ?eq .
+FILTER(!isBlank(?eq))"	Equivalent Class [NAMED ID]	owl:equivalentClass
+eq class anon	EC % SPLIT=|	optional manual							Equivalent Class [ANON ID]	
+subclass anon	SC % SPLIT=|	optional manual	'disease has feature' some cherubism	"intended for only subclass of anonymous logical expressions
+--> separated from parent id for practical purposes - ROBOT template is the same, ROBOT export differs"					SubClass Of [ANON ID]	
+subclass anon: inheritance	SC 'has material basis in' some % SPLIT=|	optional manual								
+subclass anon: anatomical location	SC 'disease has location' some % SPLIT=|	optional manual								
+subclass anon: onset	SC 'existence starts during' some % SPLIT=|	optional manual								
+subclass anon: has_material_basis_in	SC has_material_basis_in some % SPLIT=|	optional manual	autosomal dominant inheritance	do not quote standalone terms!!!						
+subclass anon: located_in	SC located_in some % SPLIT=|	optional manual			rdfs:label (preferred); IRI or CURIE (possible)					
+disjoint class	DC % SPLIT=|	optional manual						"?iri owl:disjointClass ?disjoint .
+FILTER(!isBlank(?disjoint))"	Disjoint With [NAMED ID]	owl:disjointWith
+disjoint class anon	DC % SPLIT=|	optional manual							Disjoint With [ANON ID]	
+subset(s)	AI oboInOwl:inSubset SPLIT=|	optional manual	DO_AGR_slim		any subset (aka 'slim') defined in doid-edit.owl			"?iri oboInOwl:inSubset ?subset_iri .
+?subset_iri rdfs:label ?subset ."	oboInOwl:inSubset	oboInOwl:inSubset
+alternate id(s)	A oboInOwl:hasAlternativeId SPLIT=|	optional manual	DOID:4		CURIE of deprecated term			?iri oboInOwl:hasAlternativeId ?alt_id .	oboInOwl:hasAlternativeId	oboInOwl:hasAlternativeId
+deprecated	AT owl:deprecated^^xsd:boolean	optional manual	true					?iri owl:deprecated ?deprecate .	owl:deprecated	owl:deprecated
+obsolescence reason	AI IAO:0000231	optional manual	terms merged	must be IRI or label of child of 'obsolescence reason specification' (IAO:0000225)				?iri IAO:0000231 ?obs_reason .		IAO:0000231
+term replaced by	AI IAO:0100001	optional manual	DOID:4		IRI or CURIE of term to replace by			?iri obo:IAO_0100001 ?term_replaced_by .	IAO:0100001	IAO:0100001
+consider instead	oboInOwl:consider	optional manual						?iri oboInOwl:consider ?consider .	oboInOwl:consider	oboInOwl:consider
+comment	A rdfs:comment	optional manual	This is a comment. There should only be one per term.					?iri rdfs:comment ?comment .	rdfs:comment	rdfs:comment
+created by	A oboInOwl:created_by	optional manual						?iri oboInOwl:created_by ?created_by .	oboInOwl:created_by	oboInOwl:created_by
+creation date	A oboInOwl:creation_date	optional manual						?iri oboInOwl:creation_date ?creation_date .	oboInOwl:creation_date	oboInOwl:creation_date
+obo id	A oboInOwl:id	required auto	DOID:0080943	"required data, but not necessary to include in manual curation; will be inferred from iri/curie
+
+if manually entered it must match the CURIE form of iri/curie"	OBO CURIE			?iri oboInOwl:id ?id .		oboInOwl:id
+obo namespace	A oboInOwl:hasOBONamespace	required auto	disease_ontology	"required data, but not necessary to include in manual curation; will be automatically added for any new disease
+
+if manually entered it must be ""disease_ontology"" (without quotes)"	OBO namespace of ontology: disease_ontology, symptoms, transmission_process			?iri oboInOwl:hasOBONamespace ?obo_namespace .	oboInOwl:hasOBONamespace	oboInOwl:hasOBONamespace