added control (locally and in the validation) that sample_id has to be the same in the metadata and in the FASTA header #103

2 files changed, 11 insertions, 4 deletions

diff --git a/bh20sequploader/main.py b/bh20sequploader/main.py
index 80c33c8..d2a6c5f 100644
--- a/bh20sequploader/main.py
+++ b/bh20sequploader/main.py
@@ -26,11 +26,14 @@ VALIDATED_PROJECT='lugli-j7d0g-5ct8p1i1wrgyjvp'
 
 def qc_stuff(metadata, sequence_p1, sequence_p2, do_qc=True):
     failed = False
+    sample_id = ''
     try:
         log.debug("Checking metadata" if do_qc else "Skipping metadata check")
-        if do_qc and not qc_metadata(metadata.name):
-            log.warning("Failed metadata QC")
-            failed = True
+        if do_qc:
+            sample_id = qc_metadata(metadata.name)
+            if not sample_id:
+                log.warning("Failed metadata QC")
+                failed = True
     except Exception as e:
         log.exception("Failed metadata QC")
         failed = True
@@ -46,6 +49,10 @@ def qc_stuff(metadata, sequence_p1, sequence_p2, do_qc=True):
 
             target[0] = ("reads_1."+target[0][0][6:], target[0][1])
             target[1] = ("reads_2."+target[1][0][6:], target[0][1])
+
+        if target[0][2] == 'text/fasta' and sample_id != target[0][1]:
+            raise ValueError("The sample_id field in the metadata must be the same as the FASTA header")
+
     except Exception as e:
         log.exception("Failed sequence QC")
         failed = True
diff --git a/bh20sequploader/qc_metadata.py b/bh20sequploader/qc_metadata.py
index 27657b1..87fa0b3 100644
--- a/bh20sequploader/qc_metadata.py
+++ b/bh20sequploader/qc_metadata.py
@@ -39,4 +39,4 @@ def qc_metadata(metadatafile):
     if not rslt:
         raise Exception(reason)
 
-    return True
+    return metadata['sample']['sample_id']