From 0bfe144ca40b4dfc1d75e195f3cfae7a8af94dc0 Mon Sep 17 00:00:00 2001
From: lltommy
Date: Thu, 12 Nov 2020 14:55:33 +0100
Subject: Trying to fix the submitter shape, shex and yaml

---
 bh20sequploader/bh20seq-schema.yml | 24 +++++++++++++-----------
 bh20sequploader/bh20seq-shex.rdf   | 10 +++++-----
 2 files changed, 18 insertions(+), 16 deletions(-)

diff --git a/bh20sequploader/bh20seq-schema.yml b/bh20sequploader/bh20seq-schema.yml
index 2b36e99..bc652a6 100644
--- a/bh20sequploader/bh20seq-schema.yml
+++ b/bh20sequploader/bh20seq-schema.yml
@@ -191,16 +191,16 @@ $graph:
         _id: http://purl.obolibrary.org/obo/OBI_0600047
         _type: "@id"
         noLinkCheck: true
+    assembly_method:
+      doc: Assembly method refers to how the reads were assembled into contigs for which either a de novo (http://purl.obolibrary.org/obo/GENEPIO_0001628) or mapping/reference based (http://purl.obolibrary.org/obo/GENEPIO_0002028) strategy is used.
+      type: string
+      jsonldPredicate:
+        _id: http://purl.obolibrary.org/obo/GENEPIO_0000090
     alignment_protocol:
       doc: Field for additional information on the pipeline applied to obtain the assembly
       type: string?
       jsonldPredicate:
         _id: http://www.ebi.ac.uk/efo/EFO_0004917
-    assembly_method:
-      doc: Assembly method refers to how the reads were assembled into contigs for which either a de novo (http://purl.obolibrary.org/obo/GENEPIO_0001628) or mapping/reference based (http://purl.obolibrary.org/obo/GENEPIO_0002028) strategy is used.
-      type: string?
-      jsonldPredicate:
-        _id: http://purl.obolibrary.org/obo/GENEPIO_0000090
     sequencing_coverage:
       doc: Sequence coverage defined as the average number of reads representing a given nucleotide (e.g. [100]) - if multiple technologies were used multiple float values can be submitted e.g. [100, 20]
       type: double[]?
@@ -231,23 +231,25 @@ $graph:
       jsonldPredicate:
           _id: http://semanticscience.org/resource/SIO_000172
     originating_lab:
-      doc: Name of the laboratory that took the sample
+      doc: Laboratory where the sample originates from sample
       type: string?
       jsonldPredicate:
           _id: http://purl.obolibrary.org/obo/NCIT_C37984
     lab_address:
-      doc: Address of the laboratory where the sample was taken
+      doc: Address of the laboratory
       type: string?
       jsonldPredicate:
-          _id: http://purl.obolibrary.org/obo/NCIT_C37900
-    provider_sample_id:
+          _id: http://purl.obolibrary.org/obo/NCIT_C25407
+    provider:
+      doc: Provider of the sample
       type: string?
       jsonldPredicate:
-          _id: http://www.ebi.ac.uk/efo/EFO_0001741
+          _id: http://purl.obolibrary.org/obo/NCIT_C37900
     submitter_sample_id:
+      doc: Identifer given to the sample by the submitter
       type: string?
       jsonldPredicate:
-          _id: http://purl.obolibrary.org/obo/NCIT_C42781
+          _id: http://purl.obolibrary.org/obo/NCIT_C25364
     publication:
       doc: Reference to publication of this sample (e.g. DOI, pubmed ID, ...)
       type: string?
diff --git a/bh20sequploader/bh20seq-shex.rdf b/bh20sequploader/bh20seq-shex.rdf
index 602f6a0..365a6e6 100644
--- a/bh20sequploader/bh20seq-shex.rdf
+++ b/bh20sequploader/bh20seq-shex.rdf
@@ -54,9 +54,9 @@ PREFIX wikidata: <http://www.wikidata.org/entity/>
     sio:SIO_000116 xsd:string *;        #(submitter_name), name, optional 0-*
     sio:SIO_000172 xsd:string ?;        #(submitter_address), address, optional, string
     obo:NCIT_C37984 xsd:string ?;       #(originating_lab) Laboratory, optional, string
-    obo:NCIT_C37900 xsd:string ?;       #(lab address), Provider, optional, string
-    efo:EFO_0001741 xsd:string ?;       #(provider_sample_id), submitter, optional ????????????????????
-    obo:NCIT_C42781 xsd:string ?;       #(submitter_sample_id), Author, optional string ????????????????????
+    obo:NCIT_C25407  xsd:string ?;      #(lab address), Address, optional, string
+    obo:NCIT_C37900 xsd:string ?;       #(provider), Provider, optional, string
+    obo:NCIT_C25364 xsd:string ?;       #(submitter_sample_id), identifier, optional, string
     obo:NCIT_C19026 xsd:string ?;       #(publication), Scientific Publication, optional string
     sio:SIO_000115 /https:\u002F\u002Forcid.org\u002F.{4}-.{4}-.{4}-.{4}/ {0,10};   #(submitter_orcid) identifier, optional but given regular expression
     sio:SIO_001167 xsd:string ?;        #(additional_submitter_information), comment, optional string
@@ -64,8 +64,8 @@ PREFIX wikidata: <http://www.wikidata.org/entity/>
 
 :technologyShape {
     obo:OBI_0600047 IRI {1,3} ;         #(sample_sequencing_technology), sequencing assay, optional
-    efo:EFO_0004917 xsd:string ;        #(alignment_protocol), high throughput sequence alignment protocol, optional
-    obo:GENEPIO_0000090 [obo:GENEPIO_0001628 obo:GENEPIO_0002028 ] ? ;    #(assembly method), assembly method, optional, one of the two
+    obo:GENEPIO_0000090 [obo:GENEPIO_0001628 obo:GENEPIO_0002028 ] ;    #(assembly method), assembly method, mandatory, one of the two
+    efo:EFO_0004917 xsd:string ?;        #(alignment_protocol), high throughput sequence alignment protocol, optional
     obo:FLU_0000848 xsd:double OR xsd:integer {0,3};    #(sequence coverage) sequence coverage, optional
     sio:SIO_001167 xsd:string ?;        #(additional_technology_information), comment, optional, string
 }
-- 
cgit v1.2.3