Mercurial > repos > artbio > fetch_fasta_from_ncbi

annotate fetch_fasta_from_NCBI.py @ 6:4af77e1af12a draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/fetch_fasta_from_ncbi commit 1a90ac7cdeb35399011207ab43c78043fd5c7287
author artbio
date Sat, 14 Oct 2023 22:34:56 +0000
parents 706fe8139955
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
1 #!/usr/bin/env python
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
2 # -*- coding: utf-8 -*-
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
3 """
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
4 From a taxonomy ID retrieves all the nucleotide sequences
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
5 It returns a multiFASTA nuc/prot file
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
6
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
7 Entrez Database UID common name E-utility Database Name
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
8 Nucleotide GI number nuccore
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
9 Protein GI number protein
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
10
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
11 Retrieve strategy:
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
12
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
13 esearch to get total number of UIDs (count)
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
14 esearch to get UIDs in batches
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
15 loop untile end of UIDs list:
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
16 epost to put a batch of UIDs in the history server
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
17 efetch to retrieve info from previous post
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
18
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
19 retmax of efetch is 1/10 of declared value from NCBI
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
20
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
21 queries are 1 sec delayed, to satisfy NCBI guidelines
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
22 (more than what they request)
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
23 """
3
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
24 import argparse
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
25 import http.client
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
26 import logging
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
27 import re
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
28 import sys
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
29 import time
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
30 from urllib import error, parse, request
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
31
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
32
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
33 LOG_FORMAT = '%(asctime)s|%(levelname)-8s|%(message)s'
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
34 LOG_DATEFMT = '%Y-%m-%d %H:%M:%S'
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
35 LOG_LEVELS = ['DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL']
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
36
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
37
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
38 class QueryException(Exception):
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
39 pass
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
40
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
41
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
42 class Eutils:
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
43
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
44 def __init__(self, options, logger):
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
45 """
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
46 Initialize retrieval parameters
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
47 """
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
48 self.logger = logger
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
49 self.base = "https://eutils.ncbi.nlm.nih.gov/entrez/eutils/"
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
50 self.query_string = options.query_string
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
51 self.dbname = options.dbname
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
52 if options.get_fasta:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
53 self.get_fasta = options.get_fasta
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
54 self.ids = []
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
55 self.retmax_esearch = 100000
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
56 self.retmax_efetch = 500
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
57 self.webenv = ''
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
58 self.usehistory = ''
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
59 self.query_key = ''
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
60 self.iuds_file = options.iuds_file
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
61 if self.iuds_file:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
62 with open(self.iuds_file, 'r') as f:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
63 for line in f:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
64 self.ids.append(line.rstrip())
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
65 self.count = len(self.ids) # 0 if query, some value if iuds_file
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
66
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
67 def retrieve(self):
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
68 """
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
69 Retrieve the iuds and fastas corresponding to the query
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
70 """
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
71 if len(self.ids) == 0: # retrieving from query (not required for file)
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
72 self.count = self.ecount()
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
73 # If no UIDs were found from query or file, exit
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
74 if self.count == 0:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
75 self.logger.info("found no UIDs. Exiting script.")
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
76 sys.exit(-1)
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
77 if not self.iuds_file:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
78 self.get_uids_list()
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
79 self.print_uids_list()
4
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
80 else:
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
81 # as self.ids already implemented
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
82 self.print_uids_list()
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
83 if self.get_fasta:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
84 try:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
85 self.get_sequences()
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
86 except QueryException as e:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
87 self.logger.error("Exiting script.")
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
88 raise e
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
89
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
90 def print_uids_list(self):
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
91 with open("retrieved_uid_list.txt", 'w') as f:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
92 f.write('\n'.join(self.ids))
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
93
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
94 def ecount(self):
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
95 """
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
96 just to retrieve Count (number of UIDs)
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
97 Total number of UIDs from the retrieved set to be shown in the XML
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
98 output (default=20). By default, ESearch only includes the first 20
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
99 UIDs retrieved in the XML output. If usehistory is set to 'y',
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
100 the remainder of the retrieved set will be stored on the History server
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
101 http://www.ncbi.nlm.nih.gov/books/NBK25499/#chapter4.EFetch
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
102 """
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
103 querylog = self.esearch(self.dbname, self.query_string, '', '',
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
104 'count')
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
105 self.logger.debug("Query response:")
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
106 for line in querylog:
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
107 line = line.decode('utf-8')
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
108 self.logger.debug(line.rstrip())
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
109 if '</Count>' in line:
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
110 count = int(line.split("<Count>")[1].split("</Count>")[0])
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
111 self.logger.info("Found %d UIDs" % count)
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
112 return count
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
113
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
114 def get_uids_list(self):
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
115 """
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
116 Increasing retmax allows more of the retrieved UIDs to be included in
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
117 the XML output, up to a maximum of 100,000 records.
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
118 from http://www.ncbi.nlm.nih.gov/books/NBK25499/#chapter4.ESearch
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
119 """
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
120 retmax = self.retmax_esearch
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
121 self.logger.info("retmax = %s, self.count = %s" % (retmax, self.count))
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
122 if (int(self.count) > retmax):
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
123 num_batches = int(self.count / retmax) + 1
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
124 self.usehistory = 'y'
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
125 self.logger.info("Batch size for esearch action: %d UIDs" % retmax)
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
126 self.logger.info("Number of batches for esearch action: %s"
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
127 % num_batches)
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
128 querylog = self.esearch(self.dbname, self.query_string, '', '', '')
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
129 for line in querylog:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
130 line = line.decode('utf-8')
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
131 self.logger.debug(line.rstrip())
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
132 if '<WebEnv>' in line:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
133 self.webenv = line.split("<WebEnv>")[1].split(
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
134 "</WebEnv>")[0]
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
135 self.logger.info("Will use webenv %s" % self.webenv)
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
136 for n in range(num_batches):
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
137 querylog = self.esearch(self.dbname, self.query_string,
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
138 n*retmax, retmax, '')
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
139 for line in querylog:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
140 line = line.decode('utf-8')
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
141 if '<Id>' in line and '</Id>' in line:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
142 uid = line.split("<Id>")[1].split("</Id>")[0]
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
143 self.ids.append(uid)
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
144 self.logger.info("Retrieved %d UIDs" % len(self.ids))
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
145
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
146 else:
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
147 self.logger.info("Batch size for esearch action: %d UIDs" % retmax)
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
148 self.logger.info("Number of batches for esearch action: 1")
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
149 querylog = self.esearch(self.dbname, self.query_string, 0,
3
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
150 retmax, '')
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
151 for line in querylog:
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
152 line = line.decode('utf-8')
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
153 if '<Id>' in line and '</Id>' in line:
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
154 uid = line.split("<Id>")[1].split("</Id>")[0]
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
155 self.ids.append(uid)
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
156 self.logger.info("Retrieved %d UIDs" % len(self.ids))
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
157 return self.ids
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
158
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
159 def get_sequences(self):
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
160 batch_size = self.retmax_efetch
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
161 count = self.count
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
162 uids_list = self.ids
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
163 self.logger.info("Batch size for efetch action: %d" % batch_size)
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
164 self.logger.info("Number of batches for efetch action: %d" %
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
165 ((count / batch_size) + 1))
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
166 with open(self.get_fasta, 'w') as out:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
167 for start in range(0, count, batch_size):
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
168 end = min(count, start+batch_size)
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
169 batch = uids_list[start:end]
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
170 self.logger.info("retrieving batch %d" %
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
171 ((start / batch_size) + 1))
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
172 try:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
173 mfasta = self.efetch(self.dbname, ','.join(batch))
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
174 out.write(mfasta + '\n')
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
175 except QueryException as e:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
176 self.logger.error("%s" % e.message)
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
177 raise e
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
178 request.urlcleanup()
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
179
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
180 def efetch(self, db, uid_list):
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
181 url = self.base + "efetch.fcgi"
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
182 self.logger.debug("url_efetch: %s" % url)
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
183 values = {'db': db,
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
184 'id': uid_list,
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
185 'rettype': "fasta",
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
186 'retmode': "text",
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
187 'usehistory': self.usehistory,
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
188 'WebEnv': self.webenv}
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
189 data = parse.urlencode(values)
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
190 req = request.Request(url, data.encode('utf-8'))
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
191 self.logger.debug("data: %s" % str(data))
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
192 serverTransaction = False
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
193 counter = 0
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
194 response_code = 0
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
195 while not serverTransaction:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
196 counter += 1
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
197 self.logger.info("Server Transaction Trial: %s" % (counter))
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
198 try:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
199 self.logger.debug("Going to open")
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
200 response = request.urlopen(req)
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
201 self.logger.debug("Going to get code")
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
202 response_code = response.getcode()
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
203 self.logger.debug("Going to read, de code was : %s",
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
204 str(response_code))
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
205 fasta = response.read()
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
206 self.logger.debug("Did all that")
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
207 response.close()
6
4af77e1af12a planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/fetch_fasta_from_ncbi commit 1a90ac7cdeb35399011207ab43c78043fd5c7287
artbio
parents: 5
diff changeset
208 if ((response_code != 200) or
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
209 (b"Resource temporarily unavailable" in fasta) or
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
210 (b"Error" in fasta) or (not fasta.startswith(b">"))):
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
211 serverTransaction = False
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
212 if (response_code != 200):
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
213 self.logger.info("urlopen error: Response code is not\
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
214 200")
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
215 elif ("Resource temporarily unavailable" in fasta):
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
216 self.logger.info("Ressource temporarily unavailable")
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
217 elif ("Error" in fasta):
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
218 self.logger.info("Error in fasta")
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
219 else:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
220 self.logger.info("Fasta doesn't start with '>'")
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
221 else:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
222 serverTransaction = True
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
223 except error.HTTPError as e:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
224 serverTransaction = False
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
225 self.logger.info("urlopen error:%s, %s" % (e.code, e.read()))
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
226 except error.URLError as e:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
227 serverTransaction = False
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
228 self.logger.info("urlopen error: Failed to reach a server")
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
229 self.logger.info("Reason :%s" % (e.reason))
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
230 except http.client.IncompleteRead as e:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
231 serverTransaction = False
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
232 self.logger.info("IncompleteRead error: %s" % (e.partial))
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
233 if (counter > 500):
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
234 serverTransaction = True
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
235 if (counter > 500):
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
236 raise QueryException({"message":
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
237 "500 Server Transaction Trials attempted for\
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
238 this batch. Aborting."})
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
239 fasta = self.sanitiser(self.dbname, fasta.decode('utf-8'))
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
240 time.sleep(0.1)
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
241 return fasta
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
242
3
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
243 def esearch(self, db, term, retstart, retmax, rettype):
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
244 url = self.base + "esearch.fcgi"
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
245 self.logger.debug("url: %s" % url)
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
246 values = {'db': db,
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
247 'term': term,
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
248 'rettype': rettype,
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
249 'retstart': retstart,
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
250 'retmax': retmax,
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
251 'usehistory': self.usehistory,
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
252 'WebEnv': self.webenv}
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
253 data = parse.urlencode(values)
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
254 self.logger.debug("data: %s" % str(data))
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
255 req = request.Request(url, data.encode('utf-8'))
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
256 response = request.urlopen(req)
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
257 querylog = response.readlines()
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
258 response.close()
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
259 time.sleep(1)
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
260 return querylog
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
261
4
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
262 def sanitiser(self, db, fastaseq):
6
4af77e1af12a planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/fetch_fasta_from_ncbi commit 1a90ac7cdeb35399011207ab43c78043fd5c7287
artbio
parents: 5
diff changeset
263 if (db not in "nuccore protein"):
4
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
264 return fastaseq
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
265 regex = re.compile(r"[ACDEFGHIKLMNPQRSTVWYBZ]{49,}")
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
266 sane_seqlist = []
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
267 seqlist = fastaseq.split('\n\n')
4
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
268 for seq in seqlist[:-1]:
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
269 fastalines = seq.split("\n")
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
270 if len(fastalines) < 2:
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
271 self.logger.info("Empty sequence for %s" %
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
272 ("|".join(fastalines[0].split("|")[:4])))
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
273 self.logger.info("%s download is skipped" %
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
274 ("|".join(fastalines[0].split("|")[:4])))
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
275 continue
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
276 if db == "nuccore":
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
277 badnuc = 0
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
278 for nucleotide in fastalines[1]:
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
279 if nucleotide not in "ATGC":
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
280 badnuc += 1
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
281 if float(badnuc)/len(fastalines[1]) > 0.4:
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
282 self.logger.info("%s ambiguous nucleotides in %s\
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
283 or download interrupted at this offset\
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
284 | %s" % (float(badnuc)/len(fastalines[1]),
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
285 "|".join(fastalines[0].split("|")
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
286 [:4]),
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
287 fastalines[1]))
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
288 self.logger.info("%s download is skipped" %
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
289 fastalines[0].split("|")[:4])
4
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
290 continue
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
291 """ remove spaces and trim the header to 100 chars """
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
292 fastalines[0] = fastalines[0].replace(" ", "_")[:100]
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
293 cleanseq = "\n".join(fastalines)
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
294 sane_seqlist.append(cleanseq)
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
295 elif db == "protein":
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
296 fastalines[0] = fastalines[0][0:100]
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
297 fastalines[0] = fastalines[0].replace(" ", "_")
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
298 fastalines[0] = fastalines[0].replace("[", "_")
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
299 fastalines[0] = fastalines[0].replace("]", "_")
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
300 fastalines[0] = fastalines[0].replace("=", "_")
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
301 """ because blast makedb doesn't like it """
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
302 fastalines[0] = fastalines[0].rstrip("_")
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
303 fastalines[0] = re.sub(regex, "_", fastalines[0])
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
304 cleanseq = "\n".join(fastalines)
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
305 sane_seqlist.append(cleanseq)
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
306 self.logger.info("clean sequences appended: %d" % (len(sane_seqlist)))
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
307 return "\n".join(sane_seqlist)
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
308
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
309
3
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
310 def command_parse():
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
311 parser = argparse.ArgumentParser(description='Retrieve data from NCBI')
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
312 parser.add_argument('--query', '-i', dest='query_string',
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
313 default=None, help='NCBI Query String')
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
314 parser.add_argument('--iud_file', dest='iuds_file', default=None,
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
315 help='input list of iuds to be fetched')
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
316 parser.add_argument('--dbname', '-d', dest='dbname', help='database type')
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
317 parser.add_argument('--fasta', '-F', dest='get_fasta', default=False,
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
318 help='file with retrieved fasta sequences')
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
319 parser.add_argument('--logfile', '-l', help='log file (default=stderr)')
3
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
320 parser.add_argument('--loglevel', choices=LOG_LEVELS, default='INFO',
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
321 help='logging level (default: INFO)')
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
322 args = parser.parse_args()
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
323
4
c667d0ee39f5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ca3070e85c370b914ffa0562afe12b363e05aea4
artbio
parents: 3
diff changeset
324 if args.query_string is not None and args.iuds_file is not None:
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
325 parser.error('Please choose either fetching by query (--query) \
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
326 or by uid list (--iud_file)')
3
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
327 return args
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
328
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
329
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
330 def __main__():
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
331 """ main function """
3
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
332 args = command_parse()
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
333 log_level = getattr(logging, args.loglevel)
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
334 kwargs = {'format': LOG_FORMAT,
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
335 'datefmt': LOG_DATEFMT,
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
336 'level': log_level}
3
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
337 if args.logfile:
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
338 kwargs['filename'] = args.logfile
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
339 logging.basicConfig(**kwargs)
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
340 logger = logging.getLogger('data_from_NCBI')
3
8be88084f89c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit ab45487db8cc69750f92a40d763d51ffac940e25
artbio
parents: 2
diff changeset
341 E = Eutils(args, logger)
5
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
342 try:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
343 E.retrieve()
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
344 except Exception:
706fe8139955 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit b5ef783237b244d684e26b1ed1cc333a8305ce3e"
artbio
parents: 4
diff changeset
345 sys.exit(-1)
1
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
346
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
347
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
348 if __name__ == "__main__":
7e41bbb94159 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/fetch_fasta_from_ncbi commit 6008aafac37eec1916d6b72c05d9cfcb002b8095
artbio
parents:
diff changeset
349 __main__()