From 07bc4c65535437b8e9e0744f08da8cea541d0116 Mon Sep 17 00:00:00 2001
From: Peter Amstutz
Date: Tue, 7 Apr 2020 15:28:42 -0400
Subject: Add metadata validation with schema-salad

---
 bh20sequploader/bh20seq-schema.yml | 36 ++++++++++++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)
 create mode 100644 bh20sequploader/bh20seq-schema.yml

(limited to 'bh20sequploader/bh20seq-schema.yml')

diff --git a/bh20sequploader/bh20seq-schema.yml b/bh20sequploader/bh20seq-schema.yml
new file mode 100644
index 0000000..6e0973a
--- /dev/null
+++ b/bh20sequploader/bh20seq-schema.yml
@@ -0,0 +1,36 @@
+$graph:
+
+- name: sampleInformationSchema
+  type: record
+  fields:
+    location: string
+    host: string
+    sequenceTechnology: string
+    assemblyMethod: string
+
+- name: InstituteInformationSchema
+  type: record
+  fields:
+    OriginatingLab: string
+    SubmittingLab: string
+
+- name: SubmitterInformationSchema
+  type: record
+  fields:
+    Submitter: string
+    submissionDate: string
+
+- name: VirusDetailSchema
+  type: record
+  fields:
+    VirusName: string
+    AccessionId: string
+
+- name: MainSchema
+  type: record
+  documentRoot: true
+  fields:
+    sampleInformation: sampleInformationSchema
+    InstituteInformation: InstituteInformationSchema
+    SubmitterInformation: SubmitterInformationSchema
+    VirusDetail: VirusDetailSchema
-- 
cgit v1.2.3


From d7498093d0f5e0db052ef88815d57c2648d09425 Mon Sep 17 00:00:00 2001
From: lltommy
Date: Thu, 9 Apr 2020 14:34:54 +0200
Subject: Updating schema and examples. This is still work in progress but we
 get there

---
 bh20sequploader/bh20seq-schema.yml | 60 ++++++++++++++++++++++++++------------
 example/metadata.yaml              | 49 ++++++++++++++++++++++---------
 example/minimal_example.yaml       | 14 +++++++++
 3 files changed, 91 insertions(+), 32 deletions(-)
 create mode 100644 example/minimal_example.yaml

(limited to 'bh20sequploader/bh20seq-schema.yml')

diff --git a/bh20sequploader/bh20seq-schema.yml b/bh20sequploader/bh20seq-schema.yml
index 6e0973a..38cfb48 100644
--- a/bh20sequploader/bh20seq-schema.yml
+++ b/bh20sequploader/bh20seq-schema.yml
@@ -1,36 +1,60 @@
 $graph:
 
-- name: sampleInformationSchema
+- name: hostSchema
   type: record
   fields:
-    location: string
-    host: string
-    sequenceTechnology: string
-    assemblyMethod: string
+    host_id: string
+    host_species: string
+    host_common_name: string?
+    host_sex: string?
+    host_age: int?
+    host_age_unit: string?
+    host_health_status: string?
+    host_treatment: string?
+    additional_host_information: string?
 
-- name: InstituteInformationSchema
+- name: sampleSchema
   type: record
   fields:
-    OriginatingLab: string
-    SubmittingLab: string
+    collector_name: string
+    collecting_institution: string
+    specimen_source: string?
+    collection_date: string?
+    collection_location: string?
+    sample_storage_conditions: string?
+    additional_collection_information: string?
 
-- name: SubmitterInformationSchema
+- name: virusSchema
   type: record
   fields:
-    Submitter: string
-    submissionDate: string
+    virus_species: string?
+    virus_strain: string?
 
-- name: VirusDetailSchema
+- name: technologySchema
   type: record
   fields:
-    VirusName: string
-    AccessionId: string
+    sample_sequencing_technology: string
+    sequence_assembly_method: string?
+    sequencing_coverage: string?
+
+- name: submitterSchema
+  type: record
+  fields:
+    submitter_name: string
+    submitter_address: string?
+    originating_lab: string
+    lab_address: string?
+    provider_sample_id: string?
+    submitter_sample_id: string?
+    authors: string?
+    submitter_id: string?
 
 - name: MainSchema
   type: record
   documentRoot: true
   fields:
-    sampleInformation: sampleInformationSchema
-    InstituteInformation: InstituteInformationSchema
-    SubmitterInformation: SubmitterInformationSchema
-    VirusDetail: VirusDetailSchema
+    host: hostSchema
+    sample: sampleSchema
+    virus: virusSchema?
+    technology: technologySchema
+    submitter: submitterSchema
\ No newline at end of file
diff --git a/example/metadata.yaml b/example/metadata.yaml
index 587d0be..8a93379 100644
--- a/example/metadata.yaml
+++ b/example/metadata.yaml
@@ -1,17 +1,38 @@
-sampleInformation:
-  location: "USA"
-  host : "Homo Sapiens"
-  sequenceTechnology: "Sanger"
-  assemblyMethod: "CLC Genomics"
+host:
+    host_id: XX1
+    host_species: string
+    host_common_name: string
+    host_sex: string
+    host_age: 20
+    host_age_unit: string
+    host_health_status: string
+    host_treatment: string
+    additional_host_information: string
 
-InstituteInformation:
-  OriginatingLab: "Erik's kitchen"
-  SubmittingLab: "National Institute for Viral Disease Control and Prevention, China CDC"
+sample:
+    collector_name: XXX
+    collecting_institution: XXX
+    specimen_source: XXX
+    collection_date: XXX
+    collection_location: XXX
+    sample_storage_conditions: XXX
+    additional_collection_information: XXX
 
-SubmitterInformation:
-  Submitter: "National Institute for Viral Disease Control and Prevention, China CDC"
-  submissionDate: "04-04-2020"
+virus:
+    virus_species: XX
+    virus_strain: XX
 
-VirusDetail:
-  VirusName: "hCoV-19/USA/identifer/2020"
-  AccessionId: "EPI_ISL_Random"
+technology:
+    sample_sequencing_technology: XX
+    sequence_assembly_method: XX
+    sequencing_coverage: 70x
+
+submitter:
+    submitter_name: tester
+    submitter_address: testerAdd
+    originating_lab: testLab
+    lab_address: labAdd
+    provider_sample_id: string
+    submitter_sample_id: string
+    authors: testAuthor
+    submitter_id: X12
\ No newline at end of file
diff --git a/example/minimal_example.yaml b/example/minimal_example.yaml
new file mode 100644
index 0000000..201b080
--- /dev/null
+++ b/example/minimal_example.yaml
@@ -0,0 +1,14 @@
+host:
+    host_id: XX
+    host_species: string
+
+sample:
+    collector_name: XXX
+    collecting_institution: XXX
+
+technology:
+    sample_sequencing_technology: XX
+
+submitter:
+    submitter_name: tester
+    originating_lab: testLab
\ No newline at end of file
-- 
cgit v1.2.3


From deedb2ed7046bbe81136b8d9d1edc353984d356b Mon Sep 17 00:00:00 2001
From: lltommy
Date: Thu, 9 Apr 2020 20:38:55 +0200
Subject: Adding functionality of turning keys into ontology terms (URI). This
 is work in progress - of course!

---
 bh20sequploader/bh20seq-schema.yml | 30 ++++++++++++++++++++++++------
 1 file changed, 24 insertions(+), 6 deletions(-)

(limited to 'bh20sequploader/bh20seq-schema.yml')

diff --git a/bh20sequploader/bh20seq-schema.yml b/bh20sequploader/bh20seq-schema.yml
index 38cfb48..fd9e854 100644
--- a/bh20sequploader/bh20seq-schema.yml
+++ b/bh20sequploader/bh20seq-schema.yml
@@ -3,14 +3,20 @@ $graph:
 - name: hostSchema
   type: record
   fields:
+    host_species:
+        type: string
+        jsonldPredicate:
+          _id: http://www.ebi.ac.uk/efo/EFO_0000532
     host_id: string
-    host_species: string
     host_common_name: string?
     host_sex: string?
     host_age: int?
     host_age_unit: string?
     host_health_status: string?
-    host_treatment: string?
+    host_treatment:
+      type: string?
+      jsonldPredicate:
+          _id: http://www.ebi.ac.uk/efo/EFO_0000727
     additional_host_information: string?
 
 - name: sampleSchema
@@ -20,7 +26,10 @@ $graph:
     collecting_institution: string
     specimen_source: string?
     collection_date: string?
-    collection_location: string?
+    collection_location:
+      type: string?
+      jsonldPredicate:
+        _id: https://schema.org/fromLocation
     sample_storage_conditions: string?
     additional_collection_information: string?
 
@@ -33,9 +42,18 @@ $graph:
 - name: technologySchema
   type: record
   fields:
-    sample_sequencing_technology: string
-    sequence_assembly_method: string?
-    sequencing_coverage: string?
+    sample_sequencing_technology:
+      type: string
+      jsonldPredicate:
+        _id: http://www.ebi.ac.uk/efo/EFO_0000532
+    sequence_assembly_method:
+      type: string?
+      jsonldPredicate:
+        _id: http://www.ebi.ac.uk/efo/EFO_0002699
+    sequencing_coverage:
+      type: string?
+      jsonldPredicate:
+        _id: http://purl.obolibrary.org/obo/FLU_0000848
 
 - name: submitterSchema
   type: record
-- 
cgit v1.2.3


From dbe094a150d6c969b3d69f112b3538e6a87a74a2 Mon Sep 17 00:00:00 2001
From: Peter Amstutz
Date: Thu, 9 Apr 2020 15:59:46 -0400
Subject: Add "sequencefile" for the metadata subject.

---
 bh20sequploader/bh20seq-schema.yml | 13 ++++++++++++-
 example/metadata.yaml              |  2 +-
 2 files changed, 13 insertions(+), 2 deletions(-)

(limited to 'bh20sequploader/bh20seq-schema.yml')

diff --git a/bh20sequploader/bh20seq-schema.yml b/bh20sequploader/bh20seq-schema.yml
index fd9e854..5c962d1 100644
--- a/bh20sequploader/bh20seq-schema.yml
+++ b/bh20sequploader/bh20seq-schema.yml
@@ -1,3 +1,8 @@
+$base: http://biohackathon.org/bh20-seq-schema
+$namespaces:
+  sch: https://schema.org/
+  efo: http://www.ebi.ac.uk/efo/
+  obo: http://purl.obolibrary.org/obo/
 $graph:
 
 - name: hostSchema
@@ -75,4 +80,10 @@ $graph:
     sample: sampleSchema
     virus: virusSchema?
     technology: technologySchema
-    submitter: submitterSchema
\ No newline at end of file
+    submitter: submitterSchema
+    sequencefile:
+      doc: The subject (eg the fasta/fastq file) that this metadata describes
+      type: string?
+      jsonldPredicate:
+        _id: "@id"
+        _type: "@id"
diff --git a/example/metadata.yaml b/example/metadata.yaml
index 8a93379..41ff93e 100644
--- a/example/metadata.yaml
+++ b/example/metadata.yaml
@@ -35,4 +35,4 @@ submitter:
     provider_sample_id: string
     submitter_sample_id: string
     authors: testAuthor
-    submitter_id: X12
\ No newline at end of file
+    submitter_id: X12
-- 
cgit v1.2.3


From ddabd9390d2b221786ef58a6d85200eecf82ca2f Mon Sep 17 00:00:00 2001
From: lltommy
Date: Fri, 10 Apr 2020 15:39:56 +0200
Subject: Updating schema and examples, including URIs

---
 bh20sequploader/bh20seq-schema.yml | 145 ++++++++++++++++++++++++++++++-------
 example/metadata.yaml              |   4 +
 example/minimal_example.yaml       |   6 +-
 3 files changed, 128 insertions(+), 27 deletions(-)

(limited to 'bh20sequploader/bh20seq-schema.yml')

diff --git a/bh20sequploader/bh20seq-schema.yml b/bh20sequploader/bh20seq-schema.yml
index 5c962d1..cf9b015 100644
--- a/bh20sequploader/bh20seq-schema.yml
+++ b/bh20sequploader/bh20seq-schema.yml
@@ -3,6 +3,10 @@ $namespaces:
   sch: https://schema.org/
   efo: http://www.ebi.ac.uk/efo/
   obo: http://purl.obolibrary.org/obo/
+  sio: http://semanticscience.org/resource/
+  edam: http://edamontology.org/
+  evs: http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#
+
 $graph:
 
 - name: hostSchema
@@ -12,37 +16,93 @@ $graph:
         type: string
         jsonldPredicate:
           _id: http://www.ebi.ac.uk/efo/EFO_0000532
-    host_id: string
-    host_common_name: string?
-    host_sex: string?
-    host_age: int?
-    host_age_unit: string?
-    host_health_status: string?
+    host_id:
+        type: string
+        jsonldPredicate:
+          _id: http://semanticscience.org/resource/SIO_000115
+    host_common_name:
+        type: string?
+        jsonldPredicate:
+          _id: http://purl.obolibrary.org/obo/NOMEN_0000037
+    host_sex:
+        type: string
+        jsonldPredicate:
+          _id: http://purl.obolibrary.org/obo/PATO_0000047
+    host_age:
+        type: int?
+        jsonldPredicate:
+          _id: http://purl.obolibrary.org/obo/PATO_0000011
+    host_age_unit:
+        type: string?
+        jsonldPredicate:
+          _id: http://purl.obolibrary.org/obo/UO_0000036
+    host_health_status:
+        type: string?
+        jsonldPredicate: http://purl.obolibrary.org/obo/NCIT_C25688
     host_treatment:
       type: string?
       jsonldPredicate:
           _id: http://www.ebi.ac.uk/efo/EFO_0000727
-    additional_host_information: string?
+    host_vaccination:
+      type: string?
+      jsonldPredicate:
+          _id: http://purl.obolibrary.org/obo/VO_0000001
+    additional_host_information:
+      type: string?
+      jsonldPredicate:
+          _id: http://semanticscience.org/resource/SIO_001167
 
 - name: sampleSchema
   type: record
   fields:
-    collector_name: string
-    collecting_institution: string
-    specimen_source: string?
-    collection_date: string?
+    collector_name:
+      type: string
+      jsonldPredicate:
+          _id: http://purl.obolibrary.org/obo/OBI_0001895
+    collecting_institution:
+      type: string
+      jsonldPredicate:
+          _id: http://semanticscience.org/resource/SIO_001167
+    specimen_source:
+      type: string?
+      jsonldPredicate:
+          _id: http://purl.obolibrary.org/obo/OBI_0001479
+    collection_date:
+      type: string?
+      jsonldPredicate:
+          _id: http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#C25164
     collection_location:
       type: string?
       jsonldPredicate:
-        _id: https://schema.org/fromLocation
-    sample_storage_conditions: string?
-    additional_collection_information: string?
+        _id: http://purl.obolibrary.org/obo/GAZ_00000448
+    sample_storage_conditions:
+      type: string?
+      jsonldPredicate:
+          _id: http://purl.obolibrary.org/obo/OBI_0001472
+    additional_collection_information:
+      type: string?
+      jsonldPredicate:
+          _id: http://semanticscience.org/resource/SIO_001167
+    sample_id:
+      type: string
+      jsonldPredicate:
+          _id: http://semanticscience.org/resource/SIO_000115
+    source_database_accession:
+      type: string?
+      jsonldPredicate:
+          _id: http://edamontology.org/data_2091
 
 - name: virusSchema
   type: record
   fields:
-    virus_species: string?
-    virus_strain: string?
+    virus_species:
+      type: string?
+      jsonldPredicate:
+          _id: http://edamontology.org/data_1875
+    virus_strain:
+      type: string?
+      jsonldPredicate:
+          _id: http://semanticscience.org/resource/SIO_010055
 
 - name: technologySchema
   type: record
@@ -50,7 +110,7 @@ $graph:
     sample_sequencing_technology:
       type: string
       jsonldPredicate:
-        _id: http://www.ebi.ac.uk/efo/EFO_0000532
+        _id: http://purl.obolibrary.org/obo/OBI_0600047
     sequence_assembly_method:
       type: string?
       jsonldPredicate:
@@ -63,14 +123,42 @@ $graph:
 - name: submitterSchema
   type: record
   fields:
-    submitter_name: string
-    submitter_address: string?
-    originating_lab: string
-    lab_address: string?
-    provider_sample_id: string?
-    submitter_sample_id: string?
-    authors: string?
-    submitter_id: string?
+    submitter_name:
+      type: string
+      jsonldPredicate:
+          _id: http://semanticscience.org/resource/SIO_000116
+    submitter_date:
+      type: string
+      jsonldPredicate:
+          _id: http://purl.obolibrary.org/obo/NCIT_C94162
+    submitter_address:
+      type: string?
+      jsonldPredicate:
+          _id: http://semanticscience.org/resource/SIO_000172
+    originating_lab:
+      type: string
+      jsonldPredicate:
+          _id: http://purl.obolibrary.org/obo/NCIT_C37984
+    lab_address:
+      type: string?
+      jsonldPredicate:
+          _id: http://purl.obolibrary.org/obo/OBI_0600047
+    provider_sample_id:
+      type: string?
+      jsonldPredicate:
+          _id: http://purl.obolibrary.org/obo/NCIT_C37900
+    submitter_sample_id:
+      type: string?
+      jsonldPredicate:
+          _id: http://www.ebi.ac.uk/efo/EFO_0001741
+    authors:
+      type: string?
+      jsonldPredicate:
+          _id: http://purl.obolibrary.org/obo/NCIT_C42781
+    submitter_id:
+      type: string?
+      jsonldPredicate:
+          _id: http://semanticscience.org/resource/SIO_000115
 
 - name: MainSchema
   type: record
@@ -81,9 +169,14 @@ $graph:
     virus: virusSchema?
     technology: technologySchema
     submitter: submitterSchema
+    submission:
+      type: string
+      jsonldPredicate:
+        _id: "@id"
+        #_type: "@id"
     sequencefile:
       doc: The subject (eg the fasta/fastq file) that this metadata describes
       type: string?
       jsonldPredicate:
         _id: "@id"
-        _type: "@id"
+        _type: "@id"
\ No newline at end of file
diff --git a/example/metadata.yaml b/example/metadata.yaml
index 41ff93e..a2f6e57 100644
--- a/example/metadata.yaml
+++ b/example/metadata.yaml
@@ -1,3 +1,5 @@
+submission: publicSequenceResource
+
 host:
     host_id: XX1
     host_species: string
@@ -10,6 +12,7 @@ host:
     additional_host_information: string
 
 sample:
+    sample_id: XXX
     collector_name: XXX
     collecting_institution: XXX
     specimen_source: XXX
@@ -36,3 +39,4 @@ submitter:
     submitter_sample_id: string
     authors: testAuthor
     submitter_id: X12
+    submitter_date: Subdate
diff --git a/example/minimal_example.yaml b/example/minimal_example.yaml
index 201b080..f312ab7 100644
--- a/example/minimal_example.yaml
+++ b/example/minimal_example.yaml
@@ -1,8 +1,11 @@
+submission: publicSequenceResource
+
 host:
     host_id: XX
     host_species: string
 
 sample:
+    sample_id: XXX
     collector_name: XXX
     collecting_institution: XXX
 
@@ -11,4 +14,5 @@ technology:
 
 submitter:
     submitter_name: tester
-    originating_lab: testLab
\ No newline at end of file
+    originating_lab: testLab
+    submitter_date: Subdate
\ No newline at end of file
-- 
cgit v1.2.3


From a6ba9a5203a568611a94c043fd13e2ec50f071da Mon Sep 17 00:00:00 2001
From: Peter Amstutz
Date: Fri, 10 Apr 2020 15:12:08 -0400
Subject: Validate & propagate metadata

---
 bh20seqanalyzer/main.py            | 45 ++++++++++++++++++++++++++++++++------
 bh20sequploader/bh20seq-schema.yml |  6 ++---
 bh20sequploader/main.py            |  2 +-
 bh20sequploader/qc_metadata.py     | 29 +++++++++++++++++++++++-
 4 files changed, 70 insertions(+), 12 deletions(-)

(limited to 'bh20sequploader/bh20seq-schema.yml')

diff --git a/bh20seqanalyzer/main.py b/bh20seqanalyzer/main.py
index 2030c1e..1fb51b5 100644
--- a/bh20seqanalyzer/main.py
+++ b/bh20seqanalyzer/main.py
@@ -8,6 +8,8 @@ import json
 import logging
 import ruamel.yaml
 from bh20sequploader.qc_metadata import qc_metadata
+import pkg_resources
+from schema_salad.sourceline import add_lc_filename
 
 logging.basicConfig(format="[%(asctime)s] %(levelname)s %(message)s", datefmt="%Y-%m-%d %H:%M:%S",
                     level=logging.INFO)
@@ -24,8 +26,14 @@ def validate_upload(api, collection, validated_project,
         logging.warn("Upload '%s' missing metadata.yaml", collection["name"])
         valid = False
     else:
-        metadata_content = ruamel.yaml.round_trip_load(col.open("metadata.yaml"))
-        #valid = qc_metadata(metadata_content) and valid
+        try:
+            metadata_content = ruamel.yaml.round_trip_load(col.open("metadata.yaml"))
+            metadata_content["id"] = "keep:%s/metadata.yaml" % collection["portable_data_hash"]
+            add_lc_filename(metadata_content, metadata_content["id"])
+            valid = qc_metadata(metadata_content) and valid
+        except Exception as e:
+            logging.warn(e)
+            valid = False
         if not valid:
             logging.warn("Failed metadata qc")
 
@@ -52,9 +60,10 @@ def validate_upload(api, collection, validated_project,
             "owner_uuid": validated_project,
             "name": "%s (%s)" % (collection["name"], time.asctime(time.gmtime()))}).execute()
     else:
+        pass
         # It is invalid, delete it.
-        logging.warn("Deleting '%s'" % collection["name"])
-        api.collections().delete(uuid=collection["uuid"]).execute()
+        #logging.warn("Deleting '%s'" % collection["name"])
+        #api.collections().delete(uuid=collection["uuid"]).execute()
 
     return valid
 
@@ -107,12 +116,17 @@ def start_fastq_to_fasta(api, collection,
 def start_pangenome_analysis(api,
                              analysis_project,
                              pangenome_workflow_uuid,
-                             validated_project):
+                             validated_project,
+                             schema_ref):
     validated = arvados.util.list_all(api.collections().list, filters=[["owner_uuid", "=", validated_project]])
     inputobj = {
         "inputReads": [],
         "metadata": [],
-        "subjects": []
+        "subjects": [],
+        "metadataSchema": {
+            "class": "File",
+            "location": schema_ref
+        }
     }
     for v in validated:
         inputobj["inputReads"].append({
@@ -166,12 +180,26 @@ def move_fastq_to_fasta_results(api, analysis_project, uploader_project):
             api.groups().update(uuid=p["uuid"], body={"properties": p["properties"]}).execute()
 
 
+def upload_schema(api, workflow_def_project):
+    schema_resource = pkg_resources.resource_stream('bh20sequploader.qc_metadata', "bh20seq-schema.yml")
+    c = arvados.collection.Collection()
+    with c.open("schema.yml", "wb") as f:
+        f.write(schema_resource.read())
+    pdh = c.portable_data_hash()
+    wd = api.collections().list(filters=[["owner_uuid", "=", workflow_def_project],
+                                         ["portable_data_hash", "=", pdh]]).execute()
+    if len(wd["items"]) == 0:
+        c.save_new(owner_uuid=workflow_def_project, name="Metadata schema", ensure_unique_name=True)
+    return "keep:%s/schema.yml" % pdh
+
+
 def main():
     parser = argparse.ArgumentParser(description='Analyze collections uploaded to a project')
     parser.add_argument('--uploader-project', type=str, default='lugli-j7d0g-n5clictpuvwk8aa', help='')
     parser.add_argument('--pangenome-analysis-project', type=str, default='lugli-j7d0g-y4k4uswcqi3ku56', help='')
     parser.add_argument('--fastq-project', type=str, default='lugli-j7d0g-xcjxp4oox2u1w8u', help='')
     parser.add_argument('--validated-project', type=str, default='lugli-j7d0g-5ct8p1i1wrgyjvp', help='')
+    parser.add_argument('--workflow-def-project', type=str, default='lugli-j7d0g-5hswinmpyho8dju', help='')
 
     parser.add_argument('--pangenome-workflow-uuid', type=str, default='lugli-7fd4e-mqfu9y3ofnpnho1', help='')
     parser.add_argument('--fastq-workflow-uuid', type=str, default='lugli-7fd4e-2zp9q4jo5xpif9y', help='')
@@ -183,6 +211,8 @@ def main():
 
     logging.info("Starting up, monitoring %s for uploads" % (args.uploader_project))
 
+    schema_ref = upload_schema(api, args.workflow_def_project)
+
     while True:
         move_fastq_to_fasta_results(api, args.fastq_project, args.uploader_project)
 
@@ -198,7 +228,8 @@ def main():
             start_pangenome_analysis(api,
                                      args.pangenome_analysis_project,
                                      args.pangenome_workflow_uuid,
-                                     args.validated_project)
+                                     args.validated_project,
+                                     schema_ref)
 
         copy_most_recent_result(api,
                                 args.pangenome_analysis_project,
diff --git a/bh20sequploader/bh20seq-schema.yml b/bh20sequploader/bh20seq-schema.yml
index cf9b015..a072bd7 100644
--- a/bh20sequploader/bh20seq-schema.yml
+++ b/bh20sequploader/bh20seq-schema.yml
@@ -174,9 +174,9 @@ $graph:
       jsonldPredicate:
         _id: "@id"
         #_type: "@id"
-    sequencefile:
-      doc: The subject (eg the fasta/fastq file) that this metadata describes
+    id:
+      doc: The subject (eg the fasta/fastq file) that the metadata describes
       type: string?
       jsonldPredicate:
         _id: "@id"
-        _type: "@id"
\ No newline at end of file
+        _type: "@id"
diff --git a/bh20sequploader/main.py b/bh20sequploader/main.py
index bfb8c51..2032508 100644
--- a/bh20sequploader/main.py
+++ b/bh20sequploader/main.py
@@ -22,7 +22,7 @@ def main():
 
     api = arvados.api(host=ARVADOS_API_HOST, token=ARVADOS_API_TOKEN, insecure=True)
 
-    if not qc_metadata(args.metadata.name):
+    if not bh20sequploader.qc_metadata.qc_metadata(args.metadata.name):
         print("Failed metadata qc")
         exit(1)
 
diff --git a/bh20sequploader/qc_metadata.py b/bh20sequploader/qc_metadata.py
index ebe4dfc..38edcaa 100644
--- a/bh20sequploader/qc_metadata.py
+++ b/bh20sequploader/qc_metadata.py
@@ -1,12 +1,25 @@
 import schema_salad.schema
+import schema_salad.ref_resolver
 import logging
 import pkg_resources
 import logging
+import traceback
+
+class CustomFetcher(schema_salad.ref_resolver.DefaultFetcher):
+    def check_exists(sup, url):
+        if url.startswith("keep:"):
+            return True
+        else:
+            return super().check_exists(url)
+
+    def supported_schemes(self):  # type: () -> List[str]
+        return ["file", "http", "https", "mailto", "keep"]
+
 
 def qc_metadata(metadatafile):
     schema_resource = pkg_resources.resource_stream(__name__, "bh20seq-schema.yml")
     cache = {"https://raw.githubusercontent.com/arvados/bh20-seq-resource/master/bh20sequploader/bh20seq-schema.yml": schema_resource.read().decode("utf-8")}
-    (document_loader,
+    (loader,
      avsc_names,
      schema_metadata,
      metaschema_loader) = schema_salad.schema.load_schema("https://raw.githubusercontent.com/arvados/bh20-seq-resource/master/bh20sequploader/bh20seq-schema.yml", cache=cache)
@@ -15,9 +28,23 @@ def qc_metadata(metadatafile):
         print(avsc_names)
         return False
 
+    document_loader = schema_salad.ref_resolver.Loader(
+        loader.ctx,
+        schemagraph=loader.graph,
+        foreign_properties=loader.foreign_properties,
+        idx=loader.idx,
+        cache=loader.cache,
+        fetcher_constructor=CustomFetcher,
+        skip_schemas=loader.skip_schemas,
+        url_fields=loader.url_fields,
+        allow_attachments=loader.allow_attachments,
+        session=loader.session,
+        )
+
     try:
         doc, metadata = schema_salad.schema.load_and_validate(document_loader, avsc_names, metadatafile, True)
         return True
     except Exception as e:
+        traceback.print_exc()
         logging.warn(e)
     return False
-- 
cgit v1.2.3


From ffad8c21f08aca1cf65809c398f3613846e7c8ba Mon Sep 17 00:00:00 2001
From: Peter Amstutz
Date: Fri, 10 Apr 2020 16:47:31 -0400
Subject: Propagating metadata to output works now.

Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
---
 bh20seqanalyzer/main.py            |  4 ++--
 bh20sequploader/bh20seq-schema.yml |  5 -----
 bh20sequploader/qc_metadata.py     | 26 +-------------------------
 example/metadata.yaml              |  2 --
 4 files changed, 3 insertions(+), 34 deletions(-)

(limited to 'bh20sequploader/bh20seq-schema.yml')

diff --git a/bh20seqanalyzer/main.py b/bh20seqanalyzer/main.py
index 63ff067..193a268 100644
--- a/bh20seqanalyzer/main.py
+++ b/bh20seqanalyzer/main.py
@@ -29,7 +29,7 @@ def validate_upload(api, collection, validated_project,
     else:
         try:
             metadata_content = ruamel.yaml.round_trip_load(col.open("metadata.yaml"))
-            metadata_content["id"] = "keep:%s/metadata.yaml" % collection["portable_data_hash"]
+            metadata_content["id"] = "http://arvados.org/keep:%s/metadata.yaml" % collection["portable_data_hash"]
             add_lc_filename(metadata_content, metadata_content["id"])
             valid = qc_metadata(metadata_content) and valid
         except Exception as e:
@@ -146,7 +146,7 @@ def start_pangenome_analysis(api,
             "class": "File",
             "location": "keep:%s/metadata.yaml" % v["portable_data_hash"]
         })
-        inputobj["subjects"].append("keep:%s/sequence.fasta" % v["portable_data_hash"])
+        inputobj["subjects"].append("http://arvados.org/keep:%s/sequence.fasta" % v["portable_data_hash"])
     run_workflow(api, analysis_project, pangenome_workflow_uuid, "Pangenome analysis", inputobj)
 
 
diff --git a/bh20sequploader/bh20seq-schema.yml b/bh20sequploader/bh20seq-schema.yml
index a072bd7..8a22db1 100644
--- a/bh20sequploader/bh20seq-schema.yml
+++ b/bh20sequploader/bh20seq-schema.yml
@@ -169,11 +169,6 @@ $graph:
     virus: virusSchema?
     technology: technologySchema
     submitter: submitterSchema
-    submission:
-      type: string
-      jsonldPredicate:
-        _id: "@id"
-        #_type: "@id"
     id:
       doc: The subject (eg the fasta/fastq file) that the metadata describes
       type: string?
diff --git a/bh20sequploader/qc_metadata.py b/bh20sequploader/qc_metadata.py
index 38edcaa..e477f21 100644
--- a/bh20sequploader/qc_metadata.py
+++ b/bh20sequploader/qc_metadata.py
@@ -5,21 +5,10 @@ import pkg_resources
 import logging
 import traceback
 
-class CustomFetcher(schema_salad.ref_resolver.DefaultFetcher):
-    def check_exists(sup, url):
-        if url.startswith("keep:"):
-            return True
-        else:
-            return super().check_exists(url)
-
-    def supported_schemes(self):  # type: () -> List[str]
-        return ["file", "http", "https", "mailto", "keep"]
-
-
 def qc_metadata(metadatafile):
     schema_resource = pkg_resources.resource_stream(__name__, "bh20seq-schema.yml")
     cache = {"https://raw.githubusercontent.com/arvados/bh20-seq-resource/master/bh20sequploader/bh20seq-schema.yml": schema_resource.read().decode("utf-8")}
-    (loader,
+    (document_loader,
      avsc_names,
      schema_metadata,
      metaschema_loader) = schema_salad.schema.load_schema("https://raw.githubusercontent.com/arvados/bh20-seq-resource/master/bh20sequploader/bh20seq-schema.yml", cache=cache)
@@ -28,19 +17,6 @@ def qc_metadata(metadatafile):
         print(avsc_names)
         return False
 
-    document_loader = schema_salad.ref_resolver.Loader(
-        loader.ctx,
-        schemagraph=loader.graph,
-        foreign_properties=loader.foreign_properties,
-        idx=loader.idx,
-        cache=loader.cache,
-        fetcher_constructor=CustomFetcher,
-        skip_schemas=loader.skip_schemas,
-        url_fields=loader.url_fields,
-        allow_attachments=loader.allow_attachments,
-        session=loader.session,
-        )
-
     try:
         doc, metadata = schema_salad.schema.load_and_validate(document_loader, avsc_names, metadatafile, True)
         return True
diff --git a/example/metadata.yaml b/example/metadata.yaml
index a2f6e57..c780921 100644
--- a/example/metadata.yaml
+++ b/example/metadata.yaml
@@ -1,5 +1,3 @@
-submission: publicSequenceResource
-
 host:
     host_id: XX1
     host_species: string
-- 
cgit v1.2.3


From 4f4229a61e63649730e7120c764f364078860f0f Mon Sep 17 00:00:00 2001
From: lltommy
Date: Sat, 11 Apr 2020 02:44:04 +0200
Subject: Adding descriptions to the yml schema, slight changes to the yml
 schema

---
 bh20sequploader/bh20seq-schema.yml | 45 ++++++++++++++++++++++++++++++++------
 example/metadata.yaml              |  5 +++--
 example/minimal_example.yaml       |  6 ++---
 3 files changed, 44 insertions(+), 12 deletions(-)

(limited to 'bh20sequploader/bh20seq-schema.yml')

diff --git a/bh20sequploader/bh20seq-schema.yml b/bh20sequploader/bh20seq-schema.yml
index 8a22db1..c259f16 100644
--- a/bh20sequploader/bh20seq-schema.yml
+++ b/bh20sequploader/bh20seq-schema.yml
@@ -13,41 +13,51 @@ $graph:
   type: record
   fields:
     host_species:
+        doc: Host species as defined in NCBITaxon (e.g. http://purl.obolibrary.org/obo/NCBITaxon_9606 for Homo sapiens)
         type: string
         jsonldPredicate:
           _id: http://www.ebi.ac.uk/efo/EFO_0000532
     host_id:
+        doc: Identifer for the host. If you submit multiple samples from the same host, use the same host_id for those samples
         type: string
         jsonldPredicate:
           _id: http://semanticscience.org/resource/SIO_000115
     host_common_name:
+        doc: Text label for the host species (e.g. homo sapiens)
         type: string?
         jsonldPredicate:
           _id: http://purl.obolibrary.org/obo/NOMEN_0000037
     host_sex:
+        doc: Sex of the host as define in NCIT, IRI expected (http://purl.obolibrary.org/obo/C20197 (Male), http://purl.obolibrary.org/obo/NCIT_C27993 (Female) or unkown (http://purl.obolibrary.org/obo/NCIT_C17998))
         type: string
         jsonldPredicate:
           _id: http://purl.obolibrary.org/obo/PATO_0000047
     host_age:
+        doc: Age of the host as number (e.g. 50)
         type: int?
         jsonldPredicate:
           _id: http://purl.obolibrary.org/obo/PATO_0000011
     host_age_unit:
+        doc: Unit of host age.... this field is unstable as of now (might be removed)
         type: string?
         jsonldPredicate:
           _id: http://purl.obolibrary.org/obo/UO_0000036
     host_health_status:
+        doc: A condition or state at a particular time
         type: string?
         jsonldPredicate: http://purl.obolibrary.org/obo/NCIT_C25688
     host_treatment:
+      doc: Process in which the act is intended to modify or alter
       type: string?
       jsonldPredicate:
           _id: http://www.ebi.ac.uk/efo/EFO_0000727
     host_vaccination:
+      doc: Field is unstable
       type: string?
       jsonldPredicate:
           _id: http://purl.obolibrary.org/obo/VO_0000001
     additional_host_information:
+      doc: Field for additional host information
       type: string?
       jsonldPredicate:
           _id: http://semanticscience.org/resource/SIO_001167
@@ -56,38 +66,47 @@ $graph:
   type: record
   fields:
     collector_name:
+      doc: Name of the person that took the sample
       type: string
       jsonldPredicate:
           _id: http://purl.obolibrary.org/obo/OBI_0001895
     collecting_institution:
+      doc: Institute that was responsible of sampeling
       type: string
       jsonldPredicate:
           _id: http://semanticscience.org/resource/SIO_001167
     specimen_source:
+      doc: A specimen that derives from an anatomical part or substance arising from an organism, e.g.  tissue, organ
       type: string?
       jsonldPredicate:
           _id: http://purl.obolibrary.org/obo/OBI_0001479
     collection_date:
+      doc: Date when the sample was taken
       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 Gazetteer (https://www.ebi.ac.uk/ols/ontologies/gaz) reference, e.g.  http://purl.obolibrary.org/obo/GAZ_00002845 (China)
       type: string?
       jsonldPredicate:
         _id: http://purl.obolibrary.org/obo/GAZ_00000448
     sample_storage_conditions:
+      doc: Information aboout storage of a specified type, e.g.  frozen specimen, paraffin, fresh ....
       type: string?
       jsonldPredicate:
           _id: http://purl.obolibrary.org/obo/OBI_0001472
     additional_collection_information:
+      doc: Add additional comment about the circumstances that a sample was taken
       type: string?
       jsonldPredicate:
           _id: http://semanticscience.org/resource/SIO_001167
     sample_id:
+      doc: Id of the sample as defined by the submitter
       type: string
       jsonldPredicate:
           _id: http://semanticscience.org/resource/SIO_000115
     source_database_accession:
+      doc: If data is deposit at a public resource (e.g. Genbank, ENA) enter the Accession Id here
       type: string?
       jsonldPredicate:
           _id: http://edamontology.org/data_2091
@@ -96,10 +115,12 @@ $graph:
   type: record
   fields:
     virus_species:
+      doc: The name of a taxon from the NCBI taxonomy database
       type: string?
       jsonldPredicate:
           _id: http://edamontology.org/data_1875
     virus_strain:
+      doc: Name of the virus strain
       type: string?
       jsonldPredicate:
           _id: http://semanticscience.org/resource/SIO_010055
@@ -108,14 +129,17 @@ $graph:
   type: record
   fields:
     sample_sequencing_technology:
+      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
     sequence_assembly_method:
+      doc: Protocol which provides instructions on the alignment of sequencing reads to reference genome
       type: string?
       jsonldPredicate:
         _id: http://www.ebi.ac.uk/efo/EFO_0002699
     sequencing_coverage:
+      doc: Sequence coverage defined as the average number of reads representing a given nucleotide (e.g. 100x)
       type: string?
       jsonldPredicate:
         _id: http://purl.obolibrary.org/obo/FLU_0000848
@@ -124,22 +148,22 @@ $graph:
   type: record
   fields:
     submitter_name:
+      doc: Name of the submitter
       type: string
       jsonldPredicate:
           _id: http://semanticscience.org/resource/SIO_000116
-    submitter_date:
-      type: string
-      jsonldPredicate:
-          _id: http://purl.obolibrary.org/obo/NCIT_C94162
     submitter_address:
+      doc: Address of the submitter
       type: string?
       jsonldPredicate:
           _id: http://semanticscience.org/resource/SIO_000172
     originating_lab:
+      doc: Name of the laboratory that took the sample
       type: string
       jsonldPredicate:
           _id: http://purl.obolibrary.org/obo/NCIT_C37984
     lab_address:
+      doc: Address of the laboratory where the sample was taken
       type: string?
       jsonldPredicate:
           _id: http://purl.obolibrary.org/obo/OBI_0600047
@@ -152,10 +176,17 @@ $graph:
       jsonldPredicate:
           _id: http://www.ebi.ac.uk/efo/EFO_0001741
     authors:
+      doc: Name of the author(s)
       type: string?
       jsonldPredicate:
           _id: http://purl.obolibrary.org/obo/NCIT_C42781
-    submitter_id:
+    publication:
+      doc: Reference to publication of this sample (e.g. DOI, pubmed ID, ...)
+      type: string?
+      jsonldPredicate:
+        _id: http://purl.obolibrary.org/obo/NCIT_C19026
+    submitter_orchid:
+      doc: ORCHID of the submitter
       type: string?
       jsonldPredicate:
           _id: http://semanticscience.org/resource/SIO_000115
@@ -171,7 +202,7 @@ $graph:
     submitter: submitterSchema
     id:
       doc: The subject (eg the fasta/fastq file) that the metadata describes
-      type: string?
+      type: string
       jsonldPredicate:
         _id: "@id"
-        _type: "@id"
+        #_type: "@id" # Error (contains undefined reference) if I keep this file but we need a toplevel id to tie things together
diff --git a/example/metadata.yaml b/example/metadata.yaml
index c780921..d9e8e92 100644
--- a/example/metadata.yaml
+++ b/example/metadata.yaml
@@ -1,3 +1,5 @@
+id: placeholder
+
 host:
     host_id: XX1
     host_species: string
@@ -36,5 +38,4 @@ submitter:
     provider_sample_id: string
     submitter_sample_id: string
     authors: testAuthor
-    submitter_id: X12
-    submitter_date: Subdate
+    submitter_orchid: X12
diff --git a/example/minimal_example.yaml b/example/minimal_example.yaml
index f312ab7..160d1d4 100644
--- a/example/minimal_example.yaml
+++ b/example/minimal_example.yaml
@@ -1,8 +1,9 @@
-submission: publicSequenceResource
+id: placeholder
 
 host:
     host_id: XX
     host_species: string
+    host_sex: string
 
 sample:
     sample_id: XXX
@@ -14,5 +15,4 @@ technology:
 
 submitter:
     submitter_name: tester
-    originating_lab: testLab
-    submitter_date: Subdate
\ No newline at end of file
+    originating_lab: testLab
\ No newline at end of file
-- 
cgit v1.2.3


From 31686ba2e1b2edec29229ab5adb28b366893f17b Mon Sep 17 00:00:00 2001
From: lltommy
Date: Sat, 11 Apr 2020 04:02:55 +0200
Subject: Adding noLinkCheck to root

---
 bh20sequploader/bh20seq-schema.yml | 4 +++-
 bh20sequploader/rdf-mappings.ttl   | 0
 2 files changed, 3 insertions(+), 1 deletion(-)
 create mode 100644 bh20sequploader/rdf-mappings.ttl

(limited to 'bh20sequploader/bh20seq-schema.yml')

diff --git a/bh20sequploader/bh20seq-schema.yml b/bh20sequploader/bh20seq-schema.yml
index c259f16..81a7f22 100644
--- a/bh20sequploader/bh20seq-schema.yml
+++ b/bh20sequploader/bh20seq-schema.yml
@@ -13,6 +13,7 @@ $graph:
   type: record
   fields:
     host_species:
+        ## autocomplete # NCBITAXON
         doc: Host species as defined in NCBITaxon (e.g. http://purl.obolibrary.org/obo/NCBITaxon_9606 for Homo sapiens)
         type: string
         jsonldPredicate:
@@ -205,4 +206,5 @@ $graph:
       type: string
       jsonldPredicate:
         _id: "@id"
-        #_type: "@id" # Error (contains undefined reference) if I keep this file but we need a toplevel id to tie things together
+        _type: "@id"
+        noLinkCheck: true
diff --git a/bh20sequploader/rdf-mappings.ttl b/bh20sequploader/rdf-mappings.ttl
new file mode 100644
index 0000000..e69de29
-- 
cgit v1.2.3