Mercurial > repos > devteam > blast_datatypes
annotate blast.py @ 6:a04cf51612f1 draft
Uploaded v0.0.16, MIT License, development moved to GitHub, nucleotide database definition aware of MegaBLAST index superheader
| author | peterjc |
|---|---|
| date | Mon, 23 Sep 2013 09:56:10 -0400 |
| parents | b3a3ba0c1d47 |
| children | de11e1a921c4 |
| rev | line source |
|---|---|
| 3 | 1 """ |
| 2 BlastXml class | |
| 3 """ | |
| 4 | |
| 5 from galaxy.datatypes.data import get_file_peek | |
| 4 | 6 from galaxy.datatypes.data import Text, Data |
| 3 | 7 from galaxy.datatypes.xml import GenericXml |
| 4 | 8 from galaxy.datatypes.metadata import MetadataElement |
| 3 | 9 |
|
5
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
10 |
| 3 | 11 class BlastXml( GenericXml ): |
| 12 """NCBI Blast XML Output data""" | |
| 13 file_ext = "blastxml" | |
| 14 | |
| 15 def set_peek( self, dataset, is_multi_byte=False ): | |
| 16 """Set the peek and blurb text""" | |
| 17 if not dataset.dataset.purged: | |
| 18 dataset.peek = get_file_peek( dataset.file_name, is_multi_byte=is_multi_byte ) | |
| 19 dataset.blurb = 'NCBI Blast XML data' | |
| 20 else: | |
| 21 dataset.peek = 'file does not exist' | |
| 22 dataset.blurb = 'file purged from disk' | |
| 4 | 23 |
| 3 | 24 def sniff( self, filename ): |
| 25 """ | |
| 26 Determines whether the file is blastxml | |
| 27 | |
| 28 >>> fname = get_test_fname( 'megablast_xml_parser_test1.blastxml' ) | |
| 29 >>> BlastXml().sniff( fname ) | |
| 30 True | |
| 31 >>> fname = get_test_fname( 'tblastn_four_human_vs_rhodopsin.xml' ) | |
| 32 >>> BlastXml().sniff( fname ) | |
| 33 True | |
| 34 >>> fname = get_test_fname( 'interval.interval' ) | |
| 35 >>> BlastXml().sniff( fname ) | |
| 36 False | |
| 37 """ | |
| 38 #TODO - Use a context manager on Python 2.5+ to close handle | |
| 39 handle = open(filename) | |
| 40 line = handle.readline() | |
| 41 if line.strip() != '<?xml version="1.0"?>': | |
| 42 handle.close() | |
| 43 return False | |
| 44 line = handle.readline() | |
| 45 if line.strip() not in ['<!DOCTYPE BlastOutput PUBLIC "-//NCBI//NCBI BlastOutput/EN" "http://www.ncbi.nlm.nih.gov/dtd/NCBI_BlastOutput.dtd">', | |
| 46 '<!DOCTYPE BlastOutput PUBLIC "-//NCBI//NCBI BlastOutput/EN" "NCBI_BlastOutput.dtd">']: | |
| 47 handle.close() | |
| 48 return False | |
| 49 line = handle.readline() | |
| 50 if line.strip() != '<BlastOutput>': | |
| 51 handle.close() | |
| 52 return False | |
| 53 handle.close() | |
| 54 return True | |
| 55 | |
| 56 def merge(split_files, output_file): | |
| 57 """Merging multiple XML files is non-trivial and must be done in subclasses.""" | |
| 58 if len(split_files) == 1: | |
| 59 #For one file only, use base class method (move/copy) | |
| 60 return Text.merge(split_files, output_file) | |
| 4 | 61 if not split_files: |
| 62 raise ValueError("Given no BLAST XML files, %r, to merge into %s" \ | |
| 63 % (split_files, output_file)) | |
| 3 | 64 out = open(output_file, "w") |
| 65 h = None | |
| 66 for f in split_files: | |
| 67 h = open(f) | |
| 68 body = False | |
| 69 header = h.readline() | |
| 70 if not header: | |
| 71 out.close() | |
| 72 h.close() | |
| 73 raise ValueError("BLAST XML file %s was empty" % f) | |
| 74 if header.strip() != '<?xml version="1.0"?>': | |
| 75 out.write(header) #for diagnosis | |
| 76 out.close() | |
| 77 h.close() | |
| 78 raise ValueError("%s is not an XML file!" % f) | |
| 79 line = h.readline() | |
| 80 header += line | |
| 81 if line.strip() not in ['<!DOCTYPE BlastOutput PUBLIC "-//NCBI//NCBI BlastOutput/EN" "http://www.ncbi.nlm.nih.gov/dtd/NCBI_BlastOutput.dtd">', | |
| 82 '<!DOCTYPE BlastOutput PUBLIC "-//NCBI//NCBI BlastOutput/EN" "NCBI_BlastOutput.dtd">']: | |
| 83 out.write(header) #for diagnosis | |
| 84 out.close() | |
| 85 h.close() | |
| 86 raise ValueError("%s is not a BLAST XML file!" % f) | |
| 87 while True: | |
| 88 line = h.readline() | |
| 89 if not line: | |
| 90 out.write(header) #for diagnosis | |
| 91 out.close() | |
| 92 h.close() | |
| 93 raise ValueError("BLAST XML file %s ended prematurely" % f) | |
| 94 header += line | |
| 95 if "<Iteration>" in line: | |
| 96 break | |
| 97 if len(header) > 10000: | |
| 98 #Something has gone wrong, don't load too much into memory! | |
| 99 #Write what we have to the merged file for diagnostics | |
| 100 out.write(header) | |
| 101 out.close() | |
| 102 h.close() | |
| 103 raise ValueError("BLAST XML file %s has too long a header!" % f) | |
| 104 if "<BlastOutput>" not in header: | |
| 105 out.close() | |
| 106 h.close() | |
| 107 raise ValueError("%s is not a BLAST XML file:\n%s\n..." % (f, header)) | |
| 108 if f == split_files[0]: | |
| 109 out.write(header) | |
| 110 old_header = header | |
| 111 elif old_header[:300] != header[:300]: | |
| 112 #Enough to check <BlastOutput_program> and <BlastOutput_version> match | |
| 113 out.close() | |
| 114 h.close() | |
| 115 raise ValueError("BLAST XML headers don't match for %s and %s - have:\n%s\n...\n\nAnd:\n%s\n...\n" \ | |
| 116 % (split_files[0], f, old_header[:300], header[:300])) | |
| 117 else: | |
| 118 out.write(" <Iteration>\n") | |
| 119 for line in h: | |
| 120 if "</BlastOutput_iterations>" in line: | |
| 121 break | |
| 122 #TODO - Increment <Iteration_iter-num> and if required automatic query names | |
| 123 #like <Iteration_query-ID>Query_3</Iteration_query-ID> to be increasing? | |
| 124 out.write(line) | |
| 125 h.close() | |
| 126 out.write(" </BlastOutput_iterations>\n") | |
| 127 out.write("</BlastOutput>\n") | |
| 128 out.close() | |
| 129 merge = staticmethod(merge) | |
| 130 | |
| 4 | 131 |
| 132 class _BlastDb(object): | |
| 133 """Base class for BLAST database datatype.""" | |
| 134 | |
| 135 def set_peek( self, dataset, is_multi_byte=False ): | |
| 136 """Set the peek and blurb text.""" | |
| 137 if not dataset.dataset.purged: | |
| 138 dataset.peek = "BLAST database (multiple files)" | |
| 139 dataset.blurb = "BLAST database (multiple files)" | |
| 140 else: | |
| 141 dataset.peek = 'file does not exist' | |
| 142 dataset.blurb = 'file purged from disk' | |
| 143 | |
| 144 def display_peek( self, dataset ): | |
| 145 """Create HTML content, used for displaying peek.""" | |
| 146 try: | |
| 147 return dataset.peek | |
| 148 except: | |
| 149 return "BLAST database (multiple files)" | |
| 150 | |
| 151 def display_data(self, trans, data, preview=False, filename=None, | |
| 152 to_ext=None, size=None, offset=None, **kwd): | |
| 153 """Apparently an old display method, but still gets called. | |
| 154 | |
| 155 This allows us to format the data shown in the central pane via the "eye" icon. | |
| 156 """ | |
| 157 return "This is a BLAST database." | |
| 158 | |
| 159 def get_mime(self): | |
| 160 """Returns the mime type of the datatype (pretend it is text for peek)""" | |
| 161 return 'text/plain' | |
| 162 | |
| 163 def merge(split_files, output_file): | |
| 164 """Merge BLAST databases (not implemented for now).""" | |
| 165 raise NotImplementedError("Merging BLAST databases is non-trivial (do this via makeblastdb?)") | |
| 166 | |
| 167 def split( cls, input_datasets, subdir_generator_function, split_params): | |
| 168 """Split a BLAST database (not implemented for now).""" | |
| 169 if split_params is None: | |
| 170 return None | |
| 171 raise NotImplementedError("Can't split BLAST databases") | |
| 172 | |
| 173 | |
| 174 class BlastNucDb( _BlastDb, Data ): | |
| 175 """Class for nucleotide BLAST database files.""" | |
| 176 file_ext = 'blastdbn' | |
| 177 composite_type ='basic' | |
| 178 | |
|
5
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
179 def __init__(self, **kwd): |
| 4 | 180 Data.__init__(self, **kwd) |
|
5
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
181 self.add_composite_file('blastdb.nhr', is_binary=True) # sequence headers |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
182 self.add_composite_file('blastdb.nin', is_binary=True) # index file |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
183 self.add_composite_file('blastdb.nsq', is_binary=True) # nucleotide sequences |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
184 self.add_composite_file('blastdb.nal', is_binary=False, optional=True) # alias ( -gi_mask option of makeblastdb) |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
185 self.add_composite_file('blastdb.nhd', is_binary=True, optional=True) # sorted sequence hash values ( -hash_index option of makeblastdb) |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
186 self.add_composite_file('blastdb.nhi', is_binary=True, optional=True) # index of sequence hash values ( -hash_index option of makeblastdb) |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
187 self.add_composite_file('blastdb.nnd', is_binary=True, optional=True) # sorted GI values ( -parse_seqids option of makeblastdb and gi present in the description lines) |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
188 self.add_composite_file('blastdb.nni', is_binary=True, optional=True) # index of GI values ( -parse_seqids option of makeblastdb and gi present in the description lines) |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
189 self.add_composite_file('blastdb.nog', is_binary=True, optional=True) # OID->GI lookup file ( -hash_index or -parse_seqids option of makeblastdb) |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
190 self.add_composite_file('blastdb.nsd', is_binary=True, optional=True) # sorted sequence accession values ( -hash_index or -parse_seqids option of makeblastdb) |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
191 self.add_composite_file('blastdb.nsi', is_binary=True, optional=True) # index of sequence accession values ( -hash_index or -parse_seqids option of makeblastdb) |
|
6
a04cf51612f1
Uploaded v0.0.16, MIT License, development moved to GitHub, nucleotide database definition aware of MegaBLAST index superheader
peterjc
parents:
5
diff
changeset
|
192 # self.add_composite_file('blastdb.00.idx', is_binary=True, optional=True) # first volume of the MegaBLAST index generated by makembindex |
|
a04cf51612f1
Uploaded v0.0.16, MIT License, development moved to GitHub, nucleotide database definition aware of MegaBLAST index superheader
peterjc
parents:
5
diff
changeset
|
193 # The previous line should be repeated for each index volume, with filename extensions like '.01.idx', '.02.idx', etc. |
|
a04cf51612f1
Uploaded v0.0.16, MIT License, development moved to GitHub, nucleotide database definition aware of MegaBLAST index superheader
peterjc
parents:
5
diff
changeset
|
194 self.add_composite_file('blastdb.shd', is_binary=True, optional=True) # MegaBLAST index superheader (-old_style_index false option of makembindex) |
|
5
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
195 # self.add_composite_file('blastdb.naa', is_binary=True, optional=True) # index of a WriteDB column for e.g. mask data |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
196 # self.add_composite_file('blastdb.nab', is_binary=True, optional=True) # data of a WriteDB column |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
197 # self.add_composite_file('blastdb.nac', is_binary=True, optional=True) # multiple byte order for a WriteDB column |
|
6
a04cf51612f1
Uploaded v0.0.16, MIT License, development moved to GitHub, nucleotide database definition aware of MegaBLAST index superheader
peterjc
parents:
5
diff
changeset
|
198 # The previous 3 lines should be repeated for each WriteDB column, with filename extensions like ('.nba', '.nbb', '.nbc'), ('.nca', '.ncb', '.ncc'), etc. |
| 4 | 199 |
| 200 def display_data(self, trans, data, preview=False, filename=None, | |
| 201 to_ext=None, size=None, offset=None, **kwd): | |
| 202 """Apparently an old display method, but still gets called. | |
| 203 | |
| 204 This allows us to format the data shown in the central pane via the "eye" icon. | |
| 205 """ | |
| 206 return "This is a BLAST nucleotide database." | |
| 207 | |
|
5
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
208 |
| 4 | 209 class BlastProtDb( _BlastDb, Data ): |
| 210 """Class for protein BLAST database files.""" | |
| 211 file_ext = 'blastdbp' | |
| 212 composite_type ='basic' | |
| 213 | |
|
5
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
214 def __init__(self, **kwd): |
| 4 | 215 Data.__init__(self, **kwd) |
|
5
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
216 # Component file comments are as in BlastNucDb except where noted |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
217 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
|
218 self.add_composite_file('blastdb.pin', is_binary=True) |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
219 self.add_composite_file('blastdb.psq', is_binary=True) # protein sequences |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
220 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
|
221 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
|
222 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
|
223 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
|
224 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
|
225 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
|
226 self.add_composite_file('blastdb.psi', is_binary=True, optional=True) |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
227 # self.add_composite_file('blastdb.paa', is_binary=True, optional=True) |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
228 # self.add_composite_file('blastdb.pab', is_binary=True, optional=True) |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
229 # self.add_composite_file('blastdb.pac', is_binary=True, optional=True) |
|
b3a3ba0c1d47
Uploaded v0.0.15 which updates the BLAST database definitions.
peterjc
parents:
4
diff
changeset
|
230 # The last 3 lines should be repeated for each WriteDB column, with filename extensions like ('.pba', '.pbb', '.pbc'), ('.pca', '.pcb', '.pcc'), etc. |
| 4 | 231 |
| 232 def display_data(self, trans, data, preview=False, filename=None, | |
| 233 to_ext=None, size=None, offset=None, **kwd): | |
| 234 """Apparently an old display method, but still gets called. | |
| 235 | |
| 236 This allows us to format the data shown in the central pane via the "eye" icon. | |
| 237 """ | |
| 238 return "This is a BLAST protein database." |