From 38340e0cedb465cd592ac40b11c9d22c75973fed Mon Sep 17 00:00:00 2001
From: Peter Amstutz
Date: Fri, 3 Jul 2020 21:15:48 +0000
Subject: Add --skip-qc for faster batch import

---
 bh20sequploader/main.py     | 4 ++--
 bh20sequploader/qc_fasta.py | 6 +++---
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/bh20sequploader/main.py b/bh20sequploader/main.py
index dc63bfc..cdc4c3f 100644
--- a/bh20sequploader/main.py
+++ b/bh20sequploader/main.py
@@ -24,7 +24,7 @@ UPLOAD_PROJECT='lugli-j7d0g-n5clictpuvwk8aa'
 
 def qc_stuff(metadata, sequence_p1, sequence_p2, do_qc=True):
     try:
-        log.debug("Checking metadata")
+        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")
             exit(1)
@@ -36,7 +36,7 @@ def qc_stuff(metadata, sequence_p1, sequence_p2, do_qc=True):
 
     target = []
     try:
-        log.debug("Checking FASTA/FASTQ QC")
+        log.debug("FASTA/FASTQ QC" if do_qc else "Limited FASTA/FASTQ QC")
         target.append(qc_fasta(sequence_p1, check_with_clustalw=do_qc))
         if sequence_p2:
             target.append(qc_fasta(sequence_p2))
diff --git a/bh20sequploader/qc_fasta.py b/bh20sequploader/qc_fasta.py
index 944b52c..8c6ebd3 100644
--- a/bh20sequploader/qc_fasta.py
+++ b/bh20sequploader/qc_fasta.py
@@ -54,6 +54,9 @@ def qc_fasta(arg_sequence, check_with_clustalw=True):
         sequence.seek(0)
         sequence.detach()
 
+        if not check_with_clustalw:
+            return ("sequence.fasta"+gz, seqlabel)
+
         with tempfile.NamedTemporaryFile() as tmp1:
             refstring = pkg_resources.resource_string(__name__, "SARS-CoV-2-reference.fasta")
             tmp1.write(refstring)
@@ -64,9 +67,6 @@ def qc_fasta(arg_sequence, check_with_clustalw=True):
             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")
-- 
cgit v1.2.3