Merge branch 'master' of github.com:arvados/bh20-seq-resource

3 files changed, 129 insertions, 5 deletions

diff --git a/bh20sequploader/bh20seq-schema.yml b/bh20sequploader/bh20seq-schema.yml
new file mode 100644
index 0000000..5c962d1
--- /dev/null
+++ b/bh20sequploader/bh20seq-schema.yml
@@ -0,0 +1,89 @@
+$base: http://biohackathon.org/bh20-seq-schema
+$namespaces:
+  sch: https://schema.org/
+  efo: http://www.ebi.ac.uk/efo/
+  obo: http://purl.obolibrary.org/obo/
+$graph:
+
+- name: hostSchema
+  type: record
+  fields:
+    host_species:
+        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_treatment:
+      type: string?
+      jsonldPredicate:
+          _id: http://www.ebi.ac.uk/efo/EFO_0000727
+    additional_host_information: string?
+
+- name: sampleSchema
+  type: record
+  fields:
+    collector_name: string
+    collecting_institution: string
+    specimen_source: string?
+    collection_date: string?
+    collection_location:
+      type: string?
+      jsonldPredicate:
+        _id: https://schema.org/fromLocation
+    sample_storage_conditions: string?
+    additional_collection_information: string?
+
+- name: virusSchema
+  type: record
+  fields:
+    virus_species: string?
+    virus_strain: string?
+
+- name: technologySchema
+  type: record
+  fields:
+    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
+  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:
+    host: hostSchema
+    sample: sampleSchema
+    virus: virusSchema?
+    technology: technologySchema
+    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/bh20sequploader/main.py b/bh20sequploader/main.py
index 17ad492..56cbe22 100644
--- a/bh20sequploader/main.py
+++ b/bh20sequploader/main.py
@@ -6,6 +6,7 @@ import json
 import urllib.request
 import socket
 import getpass
+from .qc_metadata import qc_metadata
 
 ARVADOS_API_HOST='lugli.arvadosapi.com'
 ARVADOS_API_TOKEN='2fbebpmbo3rw3x05ueu2i6nx70zhrsb1p22ycu3ry34m4x4462'
@@ -19,18 +20,26 @@ def main():
 
     api = arvados.api(host=ARVADOS_API_HOST, token=ARVADOS_API_TOKEN, insecure=True)
 
+    if not qc_metadata(args.metadata.name):
+        print("Failed metadata qc")
+        exit(1)
+
     col = arvados.collection.Collection(api_client=api)
 
-    print("Reading FASTA")
-    with col.open("sequence.fasta", "w") as f:
+    if args.sequence.name.endswith("fasta") or args.sequence.name.endswith("fa"):
+        target = "sequence.fasta"
+    elif args.sequence.name.endswith("fastq") or args.sequence.name.endswith("fq"):
+        target = "reads.fastq"
+
+    with col.open(target, "w") as f:
         r = args.sequence.read(65536)
         print(r[0:20])
         while r:
             f.write(r)
             r = args.sequence.read(65536)
 
-    print("Reading JSONLD")
-    with col.open("metadata.jsonld", "w") as f:
+    print("Reading metadata")
+    with col.open("metadata.yaml", "w") as f:
         r = args.metadata.read(65536)
         print(r[0:20])
         while r:
@@ -49,4 +58,7 @@ def main():
                  (properties['upload_user'], properties['upload_ip']),
                  properties=properties, ensure_unique_name=True)
 
-main()
+    print("Done")
+
+if __name__ == "__main__":
+    main()
diff --git a/bh20sequploader/qc_metadata.py b/bh20sequploader/qc_metadata.py
new file mode 100644
index 0000000..ebe4dfc
--- /dev/null
+++ b/bh20sequploader/qc_metadata.py
@@ -0,0 +1,23 @@
+import schema_salad.schema
+import logging
+import pkg_resources
+import logging
+
+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,
+     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)
+
+    if not isinstance(avsc_names, schema_salad.avro.schema.Names):
+        print(avsc_names)
+        return False
+
+    try:
+        doc, metadata = schema_salad.schema.load_and_validate(document_loader, avsc_names, metadatafile, True)
+        return True
+    except Exception as e:
+        logging.warn(e)
+    return False