annotate blast.py @ 17:3eada762af11 draft

planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
author peterjc
date Tue, 23 Oct 2018 06:24:33 -0400
parents 310ec0f47485
children 1250aab8b97a
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
1 """NCBI BLAST datatypes.
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
2
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
3 Covers the ``blastxml`` format and the BLAST databases.
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
4 """
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
5
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
6 import logging
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
7 import os
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
8 from time import sleep
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
9
15
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
10 from galaxy.datatypes.data import Data, Text, get_file_peek
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
11 from galaxy.datatypes.xml import GenericXml
9
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
12
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
13 log = logging.getLogger(__name__)
5
b3a3ba0c1d47 Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents: 4
diff changeset
14
15
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
15 # Note implicit string concatenation here to avoid excessively long lines:
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
16 _DOCTYPES = ['<!DOCTYPE BlastOutput PUBLIC "-//NCBI//NCBI BlastOutput/EN" '
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
17 '"http://www.ncbi.nlm.nih.gov/dtd/NCBI_BlastOutput.dtd">',
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
18 '<!DOCTYPE BlastOutput PUBLIC "-//NCBI//NCBI BlastOutput/EN" '
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
19 '"NCBI_BlastOutput.dtd">']
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
20
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
21
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
22 class BlastXml(GenericXml):
17
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
23 """NCBI Blast XML Output data."""
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
24
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
25 file_ext = "blastxml"
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
26
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
27 def set_peek(self, dataset, is_multi_byte=False):
17
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
28 """Set the peek and blurb text."""
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
29 if not dataset.dataset.purged:
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
30 dataset.peek = get_file_peek(dataset.file_name, is_multi_byte=is_multi_byte)
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
31 dataset.blurb = 'NCBI Blast XML data'
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
32 else:
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
33 dataset.peek = 'file does not exist'
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
34 dataset.blurb = 'file purged from disk'
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
35
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
36 def sniff(self, filename):
17
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
37 """Determine from the contents if the file is blastxml.
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
38
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
39 >>> from galaxy.datatypes.sniff import get_test_fname
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
40 >>> fname = get_test_fname('megablast_xml_parser_test1.blastxml')
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
41 >>> BlastXml().sniff(fname)
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
42 True
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
43 >>> fname = get_test_fname('tblastn_four_human_vs_rhodopsin.xml')
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
44 >>> BlastXml().sniff(fname)
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
45 True
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
46 >>> fname = get_test_fname('interval.interval')
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
47 >>> BlastXml().sniff(fname)
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
48 False
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
49 """
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
50 # TODO - Use a context manager on Python 2.5+ to close handle
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
51 handle = open(filename)
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
52 line = handle.readline()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
53 if line.strip() != '<?xml version="1.0"?>':
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
54 handle.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
55 return False
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
56 line = handle.readline()
15
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
57 if line.strip() not in _DOCTYPES:
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
58 handle.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
59 return False
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
60 line = handle.readline()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
61 if line.strip() != '<BlastOutput>':
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
62 handle.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
63 return False
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
64 handle.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
65 return True
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
66
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
67 def merge(split_files, output_file):
17
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
68 """Merge muliple BLAST XML files into one.
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
69
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
70 Merging multiple XML files is non-trivial and must be done in
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
71 subclasses, rather than the generic parent base XML class.
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
72 """
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
73 if len(split_files) == 1:
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
74 # For one file only, use base class method (move/copy)
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
75 return Text.merge(split_files, output_file)
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
76 if not split_files:
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
77 raise ValueError("Given no BLAST XML files, %r, to merge into %s"
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
78 % (split_files, output_file))
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
79 out = open(output_file, "w")
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
80 h = None
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
81 for f in split_files:
9
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
82 if not os.path.isfile(f):
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
83 log.warning("BLAST XML file %s missing, retry in 1s..." % f)
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
84 sleep(1)
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
85 if not os.path.isfile(f):
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
86 log.error("BLAST XML file %s missing" % f)
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
87 raise ValueError("BLAST XML file %s missing" % f)
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
88 h = open(f)
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
89 header = h.readline()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
90 if not header:
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
91 out.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
92 h.close()
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
93 # Retry, could be transient error with networked file system...
9
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
94 log.warning("BLAST XML file %s empty, retry in 1s..." % f)
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
95 sleep(1)
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
96 h = open(f)
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
97 header = h.readline()
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
98 if not header:
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
99 log.error("BLAST XML file %s was empty" % f)
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
100 raise ValueError("BLAST XML file %s was empty" % f)
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
101 if header.strip() != '<?xml version="1.0"?>':
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
102 out.write(header) # for diagnosis
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
103 out.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
104 h.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
105 raise ValueError("%s is not an XML file!" % f)
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
106 line = h.readline()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
107 header += line
15
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
108 if line.strip() not in _DOCTYPES:
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
109 out.write(header) # for diagnosis
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
110 out.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
111 h.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
112 raise ValueError("%s is not a BLAST XML file!" % f)
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
113 while True:
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
114 line = h.readline()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
115 if not line:
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
116 out.write(header) # for diagnosis
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
117 out.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
118 h.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
119 raise ValueError("BLAST XML file %s ended prematurely" % f)
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
120 header += line
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
121 if "<Iteration>" in line:
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
122 break
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
123 if len(header) > 10000:
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
124 # Something has gone wrong, don't load too much into memory!
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
125 # Write what we have to the merged file for diagnostics
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
126 out.write(header)
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
127 out.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
128 h.close()
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
129 raise ValueError("The header in BLAST XML file %s is too long" % f)
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
130 if "<BlastOutput>" not in header:
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
131 out.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
132 h.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
133 raise ValueError("%s is not a BLAST XML file:\n%s\n..." % (f, header))
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
134 if f == split_files[0]:
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
135 out.write(header)
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
136 old_header = header
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
137 elif old_header[:300] != header[:300]:
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
138 # Enough to check <BlastOutput_program> and <BlastOutput_version> match
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
139 out.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
140 h.close()
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
141 raise ValueError("BLAST XML headers don't match for %s and %s - have:\n%s\n...\n\nAnd:\n%s\n...\n"
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
142 % (split_files[0], f, old_header[:300], header[:300]))
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
143 else:
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
144 out.write(" <Iteration>\n")
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
145 for line in h:
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
146 if "</BlastOutput_iterations>" in line:
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
147 break
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
148 # TODO - Increment <Iteration_iter-num> and if required automatic query names
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
149 # like <Iteration_query-ID>Query_3</Iteration_query-ID> to be increasing?
3
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
150 out.write(line)
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
151 h.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
152 out.write(" </BlastOutput_iterations>\n")
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
153 out.write("</BlastOutput>\n")
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
154 out.close()
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
155 merge = staticmethod(merge)
6ef523b390e0 Uploaded correct file.
peterjc
parents:
diff changeset
156
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
157
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
158 class _BlastDb(object):
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
159 """Base class for BLAST database datatype."""
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
160
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
161 def set_peek(self, dataset, is_multi_byte=False):
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
162 """Set the peek and blurb text."""
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
163 if not dataset.dataset.purged:
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
164 dataset.peek = "BLAST database (multiple files)"
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
165 dataset.blurb = "BLAST database (multiple files)"
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
166 else:
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
167 dataset.peek = 'file does not exist'
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
168 dataset.blurb = 'file purged from disk'
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
169
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
170 def display_peek(self, dataset):
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
171 """Create HTML content, used for displaying peek."""
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
172 try:
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
173 return dataset.peek
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
174 except Exception:
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
175 return "BLAST database (multiple files)"
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
176
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
177 def display_data(self, trans, data, preview=False, filename=None,
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
178 to_ext=None, size=None, offset=None, **kwd):
17
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
179 """Documented as an old display method, but still gets called via tests etc.
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
180
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
181 This allows us to format the data shown in the central pane via the "eye" icon.
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
182 """
9
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
183 if filename is not None and filename != "index":
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
184 # Change nothing - important for the unit tests to access child files:
9
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
185 return Data.display_data(self, trans, data, preview, filename,
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
186 to_ext, size, offset, **kwd)
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
187 if self.file_ext == "blastdbn":
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
188 title = "This is a nucleotide BLAST database"
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
189 elif self.file_ext == "blastdbp":
9
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
190 title = "This is a protein BLAST database"
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
191 elif self.file_ext == "blastdbd":
10
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
192 title = "This is a domain BLAST database"
9
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
193 else:
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
194 # Error?
9
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
195 title = "This is a BLAST database."
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
196 msg = ""
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
197 try:
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
198 # Try to use any text recorded in the dummy index file:
9
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
199 handle = open(data.file_name, "rU")
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
200 msg = handle.read().strip()
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
201 handle.close()
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
202 except Exception:
9
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
203 pass
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
204 if not msg:
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
205 msg = title
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
206 # Galaxy assumes HTML for the display of composite datatypes,
9
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
207 return "<html><head><title>%s</title></head><body><pre>%s</pre></body></html>" % (title, msg)
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
208
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
209 def merge(split_files, output_file):
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
210 """Merge BLAST databases (not implemented for now)."""
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
211 raise NotImplementedError("Merging BLAST databases is non-trivial (do this via makeblastdb?)")
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
212
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
213 def split(cls, input_datasets, subdir_generator_function, split_params):
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
214 """Split a BLAST database (not implemented for now)."""
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
215 if split_params is None:
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
216 return None
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
217 raise NotImplementedError("Can't split BLAST databases")
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
218
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
219
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
220 class BlastNucDb(_BlastDb, Data):
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
221 """Class for nucleotide BLAST database files."""
17
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
222
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
223 file_ext = 'blastdbn'
9
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
224 allow_datatype_change = False
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
225 composite_type = 'basic'
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
226
5
b3a3ba0c1d47 Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents: 4
diff changeset
227 def __init__(self, **kwd):
17
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
228 """Initialize the class."""
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
229 Data.__init__(self, **kwd)
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
230 self.add_composite_file('blastdb.nhr', is_binary=True) # sequence headers
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
231 self.add_composite_file('blastdb.nin', is_binary=True) # index file
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
232 self.add_composite_file('blastdb.nsq', is_binary=True) # nucleotide sequences
15
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
233
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
234 # alias ( -gi_mask option of makeblastdb)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
235 self.add_composite_file('blastdb.nal', is_binary=False, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
236
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
237 # sorted sequence hash values ( -hash_index option of makeblastdb)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
238 self.add_composite_file('blastdb.nhd', is_binary=True, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
239
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
240 # index of sequence hash values ( -hash_index option of makeblastdb)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
241 self.add_composite_file('blastdb.nhi', is_binary=True, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
242
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
243 # sorted GI values ( -parse_seqids option of makeblastdb and gi present in the description lines)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
244 self.add_composite_file('blastdb.nnd', is_binary=True, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
245
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
246 # index of GI values ( -parse_seqids option of makeblastdb and gi present in the description lines)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
247 self.add_composite_file('blastdb.nni', is_binary=True, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
248
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
249 # OID->GI lookup file ( -hash_index or -parse_seqids option of makeblastdb)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
250 self.add_composite_file('blastdb.nog', is_binary=True, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
251
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
252 # sorted sequence accession values ( -hash_index or -parse_seqids option of makeblastdb)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
253 self.add_composite_file('blastdb.nsd', is_binary=True, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
254
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
255 # index of sequence accession values ( -hash_index or -parse_seqids option of makeblastdb)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
256 self.add_composite_file('blastdb.nsi', is_binary=True, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
257
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
258 # first volume of the MegaBLAST index generated by makembindex
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
259 # self.add_composite_file('blastdb.00.idx', is_binary=True, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
260 # The previous line should be repeated for each index volume, with filename
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
261 # extensions like '.01.idx', '.02.idx', etc.
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
262
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
263 # MegaBLAST index superheader (-old_style_index false option of makembindex)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
264 # self.add_composite_file('blastdb.shd', is_binary=True, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
265
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
266 # index of a WriteDB column for e.g. mask data
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
267 # self.add_composite_file('blastdb.naa', is_binary=True, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
268
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
269 # data of a WriteDB column
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
270 # self.add_composite_file('blastdb.nab', is_binary=True, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
271
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
272 # multiple byte order for a WriteDB column
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
273 # self.add_composite_file('blastdb.nac', is_binary=True, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
274
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
275 # The previous 3 lines should be repeated for each WriteDB column, with filename
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
276 # extensions like ('.nba', '.nbb', '.nbc'), ('.nca', '.ncb', '.ncc'), etc.
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
277
5
b3a3ba0c1d47 Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents: 4
diff changeset
278
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
279 class BlastProtDb(_BlastDb, Data):
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
280 """Class for protein BLAST database files."""
17
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
281
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
282 file_ext = 'blastdbp'
9
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
283 allow_datatype_change = False
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests
peterjc
parents: 7
diff changeset
284 composite_type = 'basic'
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
285
5
b3a3ba0c1d47 Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents: 4
diff changeset
286 def __init__(self, **kwd):
17
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
287 """Initialize the class."""
4
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support.
peterjc
parents: 3
diff changeset
288 Data.__init__(self, **kwd)
15
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
289 # Component file comments are as in BlastNucDb except where noted
5
b3a3ba0c1d47 Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents: 4
diff changeset
290 self.add_composite_file('blastdb.phr', is_binary=True)
b3a3ba0c1d47 Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents: 4
diff changeset
291 self.add_composite_file('blastdb.pin', is_binary=True)
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
292 self.add_composite_file('blastdb.psq', is_binary=True) # protein sequences
5
b3a3ba0c1d47 Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents: 4
diff changeset
293 self.add_composite_file('blastdb.phd', is_binary=True, optional=True)
b3a3ba0c1d47 Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents: 4
diff changeset
294 self.add_composite_file('blastdb.phi', is_binary=True, optional=True)
b3a3ba0c1d47 Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents: 4
diff changeset
295 self.add_composite_file('blastdb.pnd', is_binary=True, optional=True)
b3a3ba0c1d47 Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents: 4
diff changeset
296 self.add_composite_file('blastdb.pni', is_binary=True, optional=True)
b3a3ba0c1d47 Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents: 4
diff changeset
297 self.add_composite_file('blastdb.pog', is_binary=True, optional=True)
b3a3ba0c1d47 Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents: 4
diff changeset
298 self.add_composite_file('blastdb.psd', is_binary=True, optional=True)
b3a3ba0c1d47 Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents: 4
diff changeset
299 self.add_composite_file('blastdb.psi', is_binary=True, optional=True)
15
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
300 # self.add_composite_file('blastdb.paa', is_binary=True, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
301 # self.add_composite_file('blastdb.pab', is_binary=True, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
302 # self.add_composite_file('blastdb.pac', is_binary=True, optional=True)
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
303 # The last 3 lines should be repeated for each WriteDB column, with filename
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty
peterjc
parents: 14
diff changeset
304 # extensions like ('.pba', '.pbb', '.pbc'), ('.pca', '.pcb', '.pcc'), etc.
10
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
305
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
306
14
623a3fbe5340 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 3f6b5c953d522a724bbcd403bcb86f1e2757a556-dirty
peterjc
parents: 12
diff changeset
307 class BlastDomainDb(_BlastDb, Data):
10
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
308 """Class for domain BLAST database files."""
17
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
309
10
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
310 file_ext = 'blastdbd'
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
311 allow_datatype_change = False
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
312 composite_type = 'basic'
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
313
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
314 def __init__(self, **kwd):
17
3eada762af11 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit ed6325d44993c65dd9fbab02902ede0a9c0eeb80-dirty
peterjc
parents: 15
diff changeset
315 """Initialize the class."""
10
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
316 Data.__init__(self, **kwd)
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
317 self.add_composite_file('blastdb.phr', is_binary=True)
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
318 self.add_composite_file('blastdb.pin', is_binary=True)
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
319 self.add_composite_file('blastdb.psq', is_binary=True)
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
320 self.add_composite_file('blastdb.freq', is_binary=True, optional=True)
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
321 self.add_composite_file('blastdb.loo', is_binary=True, optional=True)
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
322 self.add_composite_file('blastdb.psd', is_binary=True, optional=True)
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
323 self.add_composite_file('blastdb.psi', is_binary=True, optional=True)
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
324 self.add_composite_file('blastdb.rps', is_binary=True, optional=True)
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM
peterjc
parents: 9
diff changeset
325 self.add_composite_file('blastdb.aux', is_binary=True, optional=True)