From 04df498f5cd85015afce79e1e87a3979e596dcc6 Mon Sep 17 00:00:00 2001
From: Peter Amstutz
Date: Fri, 3 Jul 2020 17:08:25 -0400
Subject: Adding --skip-qc

Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
---
 bh20sequploader/main.py     | 9 +++++----
 bh20sequploader/qc_fasta.py | 7 +++++--
 2 files changed, 10 insertions(+), 6 deletions(-)

diff --git a/bh20sequploader/main.py b/bh20sequploader/main.py
index 89c30e8..dc63bfc 100644
--- a/bh20sequploader/main.py
+++ b/bh20sequploader/main.py
@@ -22,10 +22,10 @@ ARVADOS_API_HOST='lugli.arvadosapi.com'
 ARVADOS_API_TOKEN='2fbebpmbo3rw3x05ueu2i6nx70zhrsb1p22ycu3ry34m4x4462'
 UPLOAD_PROJECT='lugli-j7d0g-n5clictpuvwk8aa'
 
-def qa_stuff(metadata, sequence_p1, sequence_p2):
+def qc_stuff(metadata, sequence_p1, sequence_p2, do_qc=True):
     try:
         log.debug("Checking metadata")
-        if not qc_metadata(metadata.name):
+        if do_qc and not qc_metadata(metadata.name):
             log.warning("Failed metadata qc")
             exit(1)
     except ValueError as e:
@@ -37,7 +37,7 @@ def qa_stuff(metadata, sequence_p1, sequence_p2):
     target = []
     try:
         log.debug("Checking FASTA/FASTQ QC")
-        target.append(qc_fasta(sequence_p1))
+        target.append(qc_fasta(sequence_p1, check_with_clustalw=do_qc))
         if sequence_p2:
             target.append(qc_fasta(sequence_p2))
             target[0] = ("reads_1."+target[0][0][6:], target[0][1])
@@ -64,11 +64,12 @@ def main():
     parser.add_argument('sequence_p1', type=argparse.FileType('rb'), help='sequence FASTA/FASTQ')
     parser.add_argument('sequence_p2', type=argparse.FileType('rb'), default=None, nargs='?', help='sequence FASTQ pair')
     parser.add_argument("--validate", action="store_true", help="Dry run, validate only")
+    parser.add_argument("--skip-qc", action="store_true", help="Skip local qc check")
     args = parser.parse_args()
 
     api = arvados.api(host=ARVADOS_API_HOST, token=ARVADOS_API_TOKEN, insecure=True)
 
-    target = qa_stuff(args.metadata, args.sequence_p1, args.sequence_p2)
+    target = qc_stuff(args.metadata, args.sequence_p1, args.sequence_p2, not args.skip_qc)
     seqlabel = target[0][1]
 
     if args.validate:
diff --git a/bh20sequploader/qc_fasta.py b/bh20sequploader/qc_fasta.py
index 1587def..944b52c 100644
--- a/bh20sequploader/qc_fasta.py
+++ b/bh20sequploader/qc_fasta.py
@@ -25,7 +25,7 @@ def read_fasta(sequence):
             raise ValueError("FASTA file contains multiple entries")
     return label, bases
 
-def qc_fasta(arg_sequence):
+def qc_fasta(arg_sequence, check_with_clustalw=True):
     log.debug("Starting qc_fasta")
     schema_resource = pkg_resources.resource_stream(__name__, "validation/formats")
     with tempfile.NamedTemporaryFile() as tmp:
@@ -64,6 +64,9 @@ def qc_fasta(arg_sequence):
             refbp = 0
             similarity = 0
             try:
+                if not check_with_clustalw:
+                    raise Exception("skipping QC")
+
                 cmd = ["clustalw", "-infile="+tmp1.name,
                        "-quicktree", "-iteration=none", "-type=DNA"]
                 print("QC checking similarity to reference")
@@ -81,7 +84,7 @@ def qc_fasta(arg_sequence):
                 print(g2.group(0))
                 print(g3.group(0))
             except Exception as e:
-                logging.warn("Error trying to QC against reference sequence using 'clustalw': %s", e)
+                logging.warn("QC against reference sequence using 'clustalw': %s", e)
 
             if refbp and (subbp/refbp) < .7:
                 raise ValueError("QC fail: submit sequence length is shorter than 70% reference")
-- 
cgit v1.2.3