aboutsummaryrefslogtreecommitdiff
path: root/bh20seqanalyzer
diff options
context:
space:
mode:
authorPeter Amstutz2020-04-10 16:20:13 -0400
committerPeter Amstutz2020-04-10 16:20:13 -0400
commitf2a3aeb6d7d8ba210d060b83ea46475eab3626ac (patch)
treead3bfe4a38739b744618618dccfe2f91ee1de769 /bh20seqanalyzer
parent1b1283131f3c684bfff2c1b165565957ac01b4be (diff)
downloadbh20-seq-resource-f2a3aeb6d7d8ba210d060b83ea46475eab3626ac.tar.gz
bh20-seq-resource-f2a3aeb6d7d8ba210d060b83ea46475eab3626ac.tar.lz
bh20-seq-resource-f2a3aeb6d7d8ba210d060b83ea46475eab3626ac.zip
Improve fasta/fastq QC
Diffstat (limited to 'bh20seqanalyzer')
-rw-r--r--bh20seqanalyzer/main.py29
1 files changed, 15 insertions, 14 deletions
diff --git a/bh20seqanalyzer/main.py b/bh20seqanalyzer/main.py
index 7626662..63ff067 100644
--- a/bh20seqanalyzer/main.py
+++ b/bh20seqanalyzer/main.py
@@ -39,20 +39,21 @@ def validate_upload(api, collection, validated_project,
logging.warn("Failed metadata qc")
if valid:
- if "sequence.fasta" in col:
- try:
- qc_fasta(col.open("sequence.fasta"))
- except Exception as e:
- logging.warn(e)
- valid = False
- else:
- if "reads.fastq" in col:
- logging.info("Upload '%s' running fastq2fasta", collection["name"])
- start_fastq_to_fasta(api, collection, fastq_project, fastq_workflow_uuid)
- return False
- else:
- valid = False
- logging.warn("Upload '%s' missing sequence.fasta", collection["name"])
+ tgt = None
+ for n in ("sequence.fasta", "reads.fastq"):
+ if n not in col:
+ continue
+ with col.open(n) as qf:
+ tgt = qc_fasta(qf)
+ if tgt != n:
+ logging.info("Expected %s but magic says it should be %s", n, tgt)
+ valid = False
+ elif tgt == "reads.fastq":
+ start_fastq_to_fasta(api, collection, fastq_project, fastq_workflow_uuid)
+ return False
+ if tgt is None:
+ valid = False
+ logging.warn("Upload '%s' does not contain sequence.fasta or reads.fastq", collection["name"])
dup = api.collections().list(filters=[["owner_uuid", "=", validated_project],
["portable_data_hash", "=", col.portable_data_hash()]]).execute()