aboutsummaryrefslogtreecommitdiff
path: root/bh20seqanalyzer
diff options
context:
space:
mode:
authorAndreaGuarracino2020-08-28 11:16:24 +0200
committerAndreaGuarracino2020-08-28 11:16:24 +0200
commit3165a31e321cbf4641f9afdcbea511ee66f673bb (patch)
tree7140a760be5903c6f5e63a38da7e0f76493530a8 /bh20seqanalyzer
parentcc8f99d50236b7d0c365990398785ecc319323ea (diff)
downloadbh20-seq-resource-3165a31e321cbf4641f9afdcbea511ee66f673bb.tar.gz
bh20-seq-resource-3165a31e321cbf4641f9afdcbea511ee66f673bb.tar.lz
bh20-seq-resource-3165a31e321cbf4641f9afdcbea511ee66f673bb.zip
added control (locally and in the validation) that sample_id has to be the same in the metadata and in the FASTA header #103
Diffstat (limited to 'bh20seqanalyzer')
-rw-r--r--bh20seqanalyzer/main.py6
1 files changed, 5 insertions, 1 deletions
diff --git a/bh20seqanalyzer/main.py b/bh20seqanalyzer/main.py
index 11349e5..5f00080 100644
--- a/bh20seqanalyzer/main.py
+++ b/bh20seqanalyzer/main.py
@@ -83,12 +83,16 @@ class SeqAnalyzer:
if n not in col:
continue
with col.open(n, 'rb') as qf:
- tgt = qc_fasta(qf)[0]
+ tgt, seqlabel, seq_type = qc_fasta(qf)
if tgt != n and tgt != paired.get(n):
errors.append("Expected %s but magic says it should be %s", n, tgt)
elif tgt in ("reads.fastq", "reads.fastq.gz", "reads_1.fastq", "reads_1.fastq.gz"):
self.start_fastq_to_fasta(collection, n, sample_id)
return False
+
+ # If it is a FASTA
+ if sample_id != seqlabel:
+ errors.append("Expected sample_id == seqlabel, but %s != %s" % (sample_id, seqlabel))
if tgt is None:
errors.append("Upload '%s' does not contain sequence.fasta, reads.fastq or reads_1.fastq", collection["name"])
except Exception as v: