Mercurial > repos > greg > ensure_synced

comparison ensure_synced.py @ 0:9180906544b6 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author greg
date Thu, 15 Aug 2019 10:37:49 -0400
parents
children 3dc919d53939
comparison
equal deleted inserted replaced
-1:000000000000 0:9180906544b6
1 #!/usr/bin/env python
2 from __future__ import print_function
3
4 import argparse
5 import psycopg2
6 import sys
7
8 from sqlalchemy import create_engine
9 from sqlalchemy import MetaData
10 from sqlalchemy.engine.url import make_url
11
12 metadata = MetaData()
13
14 SKIP_VALS = ['#CHROM', 'POS', 'ID', 'REF', 'ALT', 'QUAL', 'FILTER', 'INFO', 'FORMAT']
15
16
17 class EnsureSynced(object):
18 def __init__(self):
19 self.args = None
20 self.conn = None
21 self.parse_args()
22 self.outfh = open(self.args.output, "w")
23 self.connect_db()
24 self.engine = create_engine(self.args.database_connection_string)
25 self.metadata = MetaData(self.engine)
26 self.affy_ids_from_db = []
27 self.affy_ids_from_file = []
28
29 def connect_db(self):
30 url = make_url(self.args.database_connection_string)
31 args = url.translate_connect_args(username='user')
32 args.update(url.query)
33 assert url.get_dialect().name == 'postgresql', 'This script can only be used with PostgreSQL.'
34 self.conn = psycopg2.connect(**args)
35
36 def get_affy_ids_from_db(self):
37 cmd = "SELECT affy_id FROM sample WHERE genotype_id NOT IN (SELECT id FROM genotype WHERE coral_mlg_clonal_id = 'failed') ORDER BY affy_id;"
38 cur = self.conn.cursor()
39 cur.execute(cmd)
40 rows = cur.fetchall()
41 for row in rows:
42 self.affy_ids_from_db.append(row[0])
43
44 def get_affy_ids_from_file(self, f):
45 with open(f) as fh:
46 for line in fh:
47 line = line.strip()
48 if line in SKIP_VALS:
49 # Skip the first 9 lines in the file.
50 continue
51 self.affy_ids_from_file.append(line)
52 self.affy_ids_from_file.sort()
53
54 def get_difference(self, list1, list2):
55 if len(list1) > len(list2):
56 return list(set(list1) - set(list2))
57 return list(set(list2) - set(list1))
58
59 def log(self, msg):
60 self.outfh.write("%s\n" % msg)
61
62 def parse_args(self):
63 parser = argparse.ArgumentParser()
64 parser.add_argument('--database_connection_string', dest='database_connection_string', help='Postgres database connection string'),
65 parser.add_argument('--affy_ids_from_file', dest='affy_ids_from_file', help='Affy ids taken from all previously genotyped samples vcf file')
66 parser.add_argument('--output', dest='output', help='Output dataset'),
67 self.args = parser.parse_args()
68
69 def run(self):
70 self.get_affy_ids_from_db()
71 self.get_affy_ids_from_file(self.args.affy_ids_from_file)
72 if self.affy_ids_from_db == self.affy_ids_from_file:
73 in_sync = True
74 self.log("The selected file is in sync with the database.\n\n")
75 else:
76 in_sync = False
77 self.log("The selected file is not in sync with the database.\n\n")
78 num_affy_ids_from_db = len(self.affy_ids_from_db)
79 self.log("Number of Affymetrix ids in the database: %d\n" % num_affy_ids_from_db)
80 num_affy_ids_from_file = len(self.affy_ids_from_file)
81 self.log("Number of Affymetrix ids in the file: %d\n" % num_affy_ids_from_file)
82 if not in_sync:
83 if num_affy_ids_from_db > num_affy_ids_from_file:
84 self.log("The database contains the following Affymetrix ids that are not in the file.\n")
85 else:
86 self.log("The file contains the following Affymetrix ids that are not in the database.\n")
87 diff_list = self.get_difference(self.affy_ids_from_db, self.affy_ids_from_file)
88 for affy_id in diff_list:
89 self.log("%s\n" % affy_id)
90 self.outfh.flush()
91 self.outfh.close()
92 sys.exit(1)
93
94 def shutdown(self):
95 self.outfh.flush()
96 self.outfh.close()
97 self.conn.close()
98
99
100 if __name__ == '__main__':
101 es = EnsureSynced()
102 es.run()
103 es.shutdown()