From 535b8017ddd27a9db683f6d29368258b5c48cf5a Mon Sep 17 00:00:00 2001
From: Peter Amstutz
Date: Fri, 3 Jul 2020 20:45:17 +0000
Subject: Improving genbank import workflow

---
 scripts/docker/Dockerfile | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'scripts/docker')

diff --git a/scripts/docker/Dockerfile b/scripts/docker/Dockerfile
index 9fb33d5..8811927 100644
--- a/scripts/docker/Dockerfile
+++ b/scripts/docker/Dockerfile
@@ -7,4 +7,4 @@ RUN apt-get update && \
     libssl-dev libmagic-dev python3-magic && \
     apt-get clean
 
-RUN pip3 install bh20-seq-uploader
+RUN pip3 install bh20-seq-uploader py-dateutil
-- 
cgit v1.2.3


From 54e0ce85ef3abafb8870ef5a9244a03f16b5e3f5 Mon Sep 17 00:00:00 2001
From: Peter Amstutz
Date: Tue, 7 Jul 2020 20:41:11 +0000
Subject: minimap2 returns nothing when there is no alignment.

---
 bh20sequploader/qc_fasta.py | 15 +++++++++------
 scripts/docker/Dockerfile   |  2 +-
 2 files changed, 10 insertions(+), 7 deletions(-)

(limited to 'scripts/docker')

diff --git a/bh20sequploader/qc_fasta.py b/bh20sequploader/qc_fasta.py
index b08333e..37eb4e8 100644
--- a/bh20sequploader/qc_fasta.py
+++ b/bh20sequploader/qc_fasta.py
@@ -70,16 +70,19 @@ def qc_fasta(arg_sequence, check_with_clustalw=True):
                 similarity = 0
                 try:
                     cmd = ["minimap2", "-c", tmp1.name, tmp2.name]
-                    print("QC checking similarity to reference")
-                    print(" ".join(cmd))
+                    logging.info("QC checking similarity to reference")
+                    logging.info(" ".join(cmd))
                     result = subprocess.run(cmd, stdout=subprocess.PIPE)
+                    result.check_returncode()
                     res = result.stdout.decode("utf-8")
                     mm = res.split("\t")
-                    print(mm)
-                    # divide Number of matching bases in the mapping / Target sequence length
-                    similarity = (float(mm[9]) / float(mm[6])) * 100.0
+                    if len(mm) >= 10:
+                        # divide Number of matching bases in the mapping / Target sequence length
+                        similarity = (float(mm[9]) / float(mm[6])) * 100.0
+                    else:
+                        similarity = 0
                 except Exception as e:
-                    logging.warn("QC against reference sequence using 'minimap2': %s", e)
+                    logging.warn("QC against reference sequence using 'minimap2': %s", e, exc_info=e)
 
                 if similarity and similarity < 70.0:
                     raise ValueError("QC fail: alignment to reference was less than 70%% (was %2.2f%%)" % (similarity))
diff --git a/scripts/docker/Dockerfile b/scripts/docker/Dockerfile
index 8811927..02829d4 100644
--- a/scripts/docker/Dockerfile
+++ b/scripts/docker/Dockerfile
@@ -3,7 +3,7 @@ FROM debian:10
 RUN apt-get update && \
     apt-get -yq --no-install-recommends -o Acquire::Retries=6 install \
     python3 python3-pip python3-setuptools python3-dev python-pycurl \
-    clustalw python3-biopython libcurl4-openssl-dev build-essential \
+    minimap2 python3-biopython libcurl4-openssl-dev build-essential \
     libssl-dev libmagic-dev python3-magic && \
     apt-get clean
 
-- 
cgit v1.2.3