aboutsummaryrefslogtreecommitdiff
path: root/bh20sequploader/main.py
diff options
context:
space:
mode:
authorPjotr Prins2020-04-12 12:49:06 -0500
committerGitHub2020-04-12 12:49:06 -0500
commit23722a63682fdffe51efca55b40573fa27370973 (patch)
tree7628ccda2671fe9a3e9ff204918c965d4bb9e257 /bh20sequploader/main.py
parent10ccb97cab69cb704c154387d544a74cd38d3cdf (diff)
parent92e1608b2d8b21f2001d7bf480301d314337fdc0 (diff)
downloadbh20-seq-resource-23722a63682fdffe51efca55b40573fa27370973.tar.gz
bh20-seq-resource-23722a63682fdffe51efca55b40573fa27370973.tar.lz
bh20-seq-resource-23722a63682fdffe51efca55b40573fa27370973.zip
Merge branch 'master' into patch-1
Diffstat (limited to 'bh20sequploader/main.py')
-rw-r--r--bh20sequploader/main.py26
1 files changed, 20 insertions, 6 deletions
diff --git a/bh20sequploader/main.py b/bh20sequploader/main.py
index 17ad492..4a225f6 100644
--- a/bh20sequploader/main.py
+++ b/bh20sequploader/main.py
@@ -3,9 +3,13 @@ import time
import arvados
import arvados.collection
import json
+import magic
+from pathlib import Path
import urllib.request
import socket
import getpass
+from .qc_metadata import qc_metadata
+from .qc_fasta import qc_fasta
ARVADOS_API_HOST='lugli.arvadosapi.com'
ARVADOS_API_TOKEN='2fbebpmbo3rw3x05ueu2i6nx70zhrsb1p22ycu3ry34m4x4462'
@@ -13,29 +17,36 @@ UPLOAD_PROJECT='lugli-j7d0g-n5clictpuvwk8aa'
def main():
parser = argparse.ArgumentParser(description='Upload SARS-CoV-19 sequences for analysis')
- parser.add_argument('sequence', type=argparse.FileType('r'), help='sequence FASTA')
+ parser.add_argument('sequence', type=argparse.FileType('r'), help='sequence FASTA/FASTQ')
parser.add_argument('metadata', type=argparse.FileType('r'), help='sequence metadata json')
args = parser.parse_args()
api = arvados.api(host=ARVADOS_API_HOST, token=ARVADOS_API_TOKEN, insecure=True)
+ target = qc_fasta(args.sequence)
+
+ if not qc_metadata(args.metadata.name):
+ print("Failed metadata qc")
+ exit(1)
+
col = arvados.collection.Collection(api_client=api)
- print("Reading FASTA")
- with col.open("sequence.fasta", "w") as f:
+ with col.open(target, "w") as f:
r = args.sequence.read(65536)
print(r[0:20])
while r:
f.write(r)
r = args.sequence.read(65536)
+ args.sequence.close()
- print("Reading JSONLD")
- with col.open("metadata.jsonld", "w") as f:
+ print("Reading metadata")
+ with col.open("metadata.yaml", "w") as f:
r = args.metadata.read(65536)
print(r[0:20])
while r:
f.write(r)
r = args.metadata.read(65536)
+ args.metadata.close()
external_ip = urllib.request.urlopen('https://ident.me').read().decode('utf8')
@@ -49,4 +60,7 @@ def main():
(properties['upload_user'], properties['upload_ip']),
properties=properties, ensure_unique_name=True)
-main()
+ print("Done")
+
+if __name__ == "__main__":
+ main()