blast_datatypes: blast.py annotate

annotate blast.py @ 15:310ec0f47485 draft

planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty

author	peterjc
date	Thu, 09 Feb 2017 11:16:00 -0500
parents	623a3fbe5340
children	3eada762af11

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):
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	23 """NCBI Blast XML Output data"""
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	24 file_ext = "blastxml"
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	25
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	26 def set_peek(self, dataset, is_multi_byte=False):
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	27 """Set the peek and blurb text"""
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	28 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	29 dataset.peek = get_file_peek(dataset.file_name, is_multi_byte=is_multi_byte)
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	30 dataset.blurb = 'NCBI Blast XML data'
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	31 else:
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	32 dataset.peek = 'file does not exist'
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	33 dataset.blurb = 'file purged from disk'
4 f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	34
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	35 def sniff(self, filename):
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 """Determines whether the file is 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	37
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 >>> 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	39 >>> 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	40 >>> BlastXml().sniff(fname)
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	41 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	42 >>> 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	43 >>> BlastXml().sniff(fname)
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	44 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	45 >>> 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	46 >>> BlastXml().sniff(fname)
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	47 False
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	48 """
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	49 # TODO - Use a context manager on Python 2.5+ to close handle
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	50 handle = open(filename)
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	51 line = handle.readline()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	52 if line.strip() != '<?xml version="1.0"?>':
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	53 handle.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	54 return False
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	55 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	56 if line.strip() not in _DOCTYPES:
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	57 handle.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	58 return False
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	59 line = handle.readline()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	60 if line.strip() != '<BlastOutput>':
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	61 handle.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	62 return False
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	63 handle.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	64 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	65
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	66 def merge(split_files, output_file):
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	67 """Merging multiple XML files is non-trivial and must be done in subclasses."""
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	68 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	69 # For one file only, use base class method (move/copy)
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	70 return Text.merge(split_files, output_file)
4 f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	71 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	72 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	73 % (split_files, output_file))
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	74 out = open(output_file, "w")
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	75 h = None
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	76 for f in split_files:
9 e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	77 if not os.path.isfile(f):
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	78 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	79 sleep(1)
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	80 if not os.path.isfile(f):
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	81 log.error("BLAST XML file %s missing" % f)
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	82 raise ValueError("BLAST XML file %s missing" % f)
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	83 h = open(f)
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	84 header = h.readline()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	85 if not header:
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	86 out.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	87 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	88 # 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	89 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	90 sleep(1)
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	91 h = open(f)
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	92 header = h.readline()
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	93 if not header:
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	94 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	95 raise ValueError("BLAST XML file %s was empty" % f)
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	96 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	97 out.write(header) # for diagnosis
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	98 out.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	99 h.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	100 raise ValueError("%s is not an XML file!" % f)
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	101 line = h.readline()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	102 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	103 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	104 out.write(header) # for diagnosis
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	105 out.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	106 h.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	107 raise ValueError("%s is not a BLAST XML file!" % f)
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	108 while True:
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	109 line = h.readline()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	110 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	111 out.write(header) # for diagnosis
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	112 out.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	113 h.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	114 raise ValueError("BLAST XML file %s ended prematurely" % f)
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	115 header += line
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	116 if "<Iteration>" in line:
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	117 break
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	118 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	119 # 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	120 # Write what we have to the merged file for diagnostics
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	121 out.write(header)
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	122 out.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	123 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	124 raise ValueError("The header in BLAST XML file %s is too long" % f)
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	125 if "<BlastOutput>" not in header:
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	126 out.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	127 h.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	128 raise ValueError("%s is not a BLAST XML file:\n%s\n..." % (f, header))
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	129 if f == split_files[0]:
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	130 out.write(header)
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	131 old_header = header
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	132 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	133 # Enough to check <BlastOutput_program> and <BlastOutput_version> match
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	134 out.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	135 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	136 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	137 % (split_files[0], f, old_header[:300], header[:300]))
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	138 else:
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	139 out.write(" <Iteration>\n")
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	140 for line in h:
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	141 if "</BlastOutput_iterations>" in line:
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	142 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	143 # 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	144 # like <Iteration_query-ID>Query_3</Iteration_query-ID> to be increasing?
3 6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	145 out.write(line)
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	146 h.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	147 out.write(" </BlastOutput_iterations>\n")
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	148 out.write("</BlastOutput>\n")
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	149 out.close()
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	150 merge = staticmethod(merge)
6ef523b390e0 Uploaded correct file. peterjc parents: diff changeset	151
4 f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	152
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	153 class _BlastDb(object):
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	154 """Base class for BLAST database datatype."""
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	155
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	156 def set_peek(self, dataset, is_multi_byte=False):
4 f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	157 """Set the peek and blurb text."""
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	158 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	159 dataset.peek = "BLAST database (multiple files)"
4 f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	160 dataset.blurb = "BLAST database (multiple files)"
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	161 else:
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	162 dataset.peek = 'file does not exist'
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	163 dataset.blurb = 'file purged from disk'
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	164
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	165 def display_peek(self, dataset):
4 f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	166 """Create HTML content, used for displaying peek."""
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	167 try:
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	168 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	169 except Exception:
4 f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	170 return "BLAST database (multiple files)"
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	171
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	172 def display_data(self, trans, data, preview=False, filename=None,
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	173 to_ext=None, size=None, offset=None, **kwd):
9 e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	174 """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	175
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	176 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	177 """
9 e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	178 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	179 # 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	180 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	181 to_ext, size, offset, **kwd)
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	182 if self.file_ext == "blastdbn":
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	183 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	184 elif self.file_ext == "blastdbp":
9 e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	185 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	186 elif self.file_ext == "blastdbd":
10 939a600f45e9 Uploaded v0.0.19, domain databases and PSSM peterjc parents: 9 diff changeset	187 title = "This is a domain BLAST database"
9 e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	188 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	189 # Error?
9 e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	190 title = "This is a BLAST database."
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	191 msg = ""
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	192 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	193 # 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	194 handle = open(data.file_name, "rU")
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	195 msg = handle.read().strip()
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	196 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	197 except Exception:
9 e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	198 pass
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	199 if not msg:
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	200 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	201 # 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	202 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	203
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	204 def merge(split_files, output_file):
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	205 """Merge BLAST databases (not implemented for now)."""
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	206 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	207
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	208 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	209 """Split a BLAST database (not implemented for now)."""
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	210 if split_params is None:
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	211 return None
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	212 raise NotImplementedError("Can't split BLAST databases")
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	213
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	214
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	215 class BlastNucDb(_BlastDb, Data):
4 f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	216 """Class for nucleotide BLAST database files."""
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	217 file_ext = 'blastdbn'
9 e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	218 allow_datatype_change = False
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	219 composite_type = 'basic'
4 f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	220
5 b3a3ba0c1d47 Uploaded v0.0.15 which updates the BLAST database definitions. peterjc parents: 4 diff changeset	221 def __init__(self, **kwd):
4 f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	222 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	223 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	224 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	225 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	226
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty peterjc parents: 14 diff changeset	227 # 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	228 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	229
310ec0f47485 planemo upload for repository https://github.com/peterjc/galaxy_blast/tree/master/datatypes/blast_datatypes/ commit 96d587fc6f6cab23c597e88a83daf7eecd0d4162-dirty peterjc parents: 14 diff changeset	230 # 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	231 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	232
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 # 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	234 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	235
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 # 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	237 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	238
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 # 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	240 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	241
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 # 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	243 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	244
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 # 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	246 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	247
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 # 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	249 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	250
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 # 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	252 # 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	253 # 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	254 # 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	255
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 # 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	257 # 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	258
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 # 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	260 # 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	261
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 # 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	263 # 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	264
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 # 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	266 # 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	267
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 # 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	269 # extensions like ('.nba', '.nbb', '.nbc'), ('.nca', '.ncb', '.ncc'), etc.
4 f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	270
5 b3a3ba0c1d47 Uploaded v0.0.15 which updates the BLAST database definitions. peterjc parents: 4 diff changeset	271
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	272 class BlastProtDb(_BlastDb, Data):
4 f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	273 """Class for protein BLAST database files."""
f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	274 file_ext = 'blastdbp'
9 e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	275 allow_datatype_change = False
e36c60d13c94 Uploaded v0.0.18, tweak display_data for running tests peterjc parents: 7 diff changeset	276 composite_type = 'basic'
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 def __init__(self, **kwd):
4 f9a7783ed7b6 Uploaded v0.0.14 adding BLAST database support. peterjc parents: 3 diff changeset	279 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	280 # 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	281 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	282 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	283 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	284 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	285 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	286 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	287 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	288 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	289 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	290 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	291 # 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	292 # 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	293 # 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	294 # 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	295 # extensions like ('.pba', '.pbb', '.pbc'), ('.pca', '.pcb', '.pcc'), etc.
10 939a600f45e9 Uploaded v0.0.19, domain databases and PSSM peterjc parents: 9 diff changeset	296
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM peterjc parents: 9 diff changeset	297
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	298 class BlastDomainDb(_BlastDb, Data):
10 939a600f45e9 Uploaded v0.0.19, domain databases and PSSM peterjc parents: 9 diff changeset	299 """Class for domain BLAST database files."""
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM peterjc parents: 9 diff changeset	300 file_ext = 'blastdbd'
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM peterjc parents: 9 diff changeset	301 allow_datatype_change = False
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM peterjc parents: 9 diff changeset	302 composite_type = 'basic'
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM peterjc parents: 9 diff changeset	303
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM peterjc parents: 9 diff changeset	304 def __init__(self, **kwd):
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM peterjc parents: 9 diff changeset	305 Data.__init__(self, **kwd)
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM peterjc parents: 9 diff changeset	306 self.add_composite_file('blastdb.phr', is_binary=True)
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM peterjc parents: 9 diff changeset	307 self.add_composite_file('blastdb.pin', is_binary=True)
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM peterjc parents: 9 diff changeset	308 self.add_composite_file('blastdb.psq', is_binary=True)
939a600f45e9 Uploaded v0.0.19, domain databases and PSSM peterjc parents: 9 diff changeset	309 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	310 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	311 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	312 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	313 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	314 self.add_composite_file('blastdb.aux', is_binary=True, optional=True)

Mercurial > repos > devteam > blast_datatypes

annotate blast.py @ 15:310ec0f47485 draft