allpy: allpy/base.py annotate

allpy

annotate allpy/base.py @ 421:42a05616a09b

merge

author	boris <bnagaev@gmail.com>
date	Mon, 14 Feb 2011 12:07:08 +0300
parents	a4d7438c142f 438e1951f559
children	8cd65dc65e99

rev	line source
me@261	1 import sys
bnagaev@357	2 import re
me@261	3
me@315	4 import util
bnagaev@418	5 import fileio
me@260	6
dendik@382	7 # import this very module as means of having all related classes in one place
dendik@382	8 import base
dendik@382	9
me@306	10 default_gaps = set((".", "-", "~"))
me@306	11 """Set of characters to recoginze as gaps when parsing alignment."""
me@306	12
me@328	13 class Monomer(object):
me@328	14 """Monomer object."""
me@260	15
me@328	16 type = None
me@328	17 """Either of 'dna', 'rna', 'protein'."""
me@260	18
dendik@382	19 types = base
dendik@382	20 """Mapping of related types. SHOULD be redefined in subclasses."""
dendik@382	21
me@260	22 by_code1 = {}
me@328	23 """A mapping from 1-letter code to Monomer subclass."""
me@328	24
me@260	25 by_code3 = {}
me@328	26 """A mapping from 3-letter code to Monomer subclass."""
me@328	27
me@260	28 by_name = {}
me@328	29 """A mapping from full monomer name to Monomer subclass."""
me@260	30
me@260	31 @classmethod
me@328	32 def _subclass(cls, name='', code1='', code3='', is_modified=False):
me@328	33 """Create new subclass of Monomer for given monomer type."""
me@328	34 class TheMonomer(cls):
me@328	35 pass
me@328	36 name = name.strip().capitalize()
me@328	37 code1 = code1.upper()
me@328	38 code3 = code3.upper()
bnagaev@357	39 TheMonomer.__name__ = re.sub(r"[^\w]", "_", name)
me@328	40 TheMonomer.name = name
me@328	41 TheMonomer.code1 = code1
me@328	42 TheMonomer.code3 = code3
me@328	43 TheMonomer.is_modified = is_modified
me@328	44 if not is_modified:
me@328	45 cls.by_code1[code1] = TheMonomer
me@328	46 cls.by_code3[code3] = TheMonomer
me@328	47 cls.by_name[name] = TheMonomer
me@328	48 # We duplicate distinguished long names into Monomer itself, so that we
me@328	49 # can use Monomer.from_code3 to create the relevant type of monomer.
me@328	50 Monomer.by_code3[code3] = TheMonomer
me@328	51 Monomer.by_name[name] = TheMonomer
me@260	52
me@328	53 @classmethod
me@353	54 def _initialize(cls, codes=None):
me@328	55 """Create all relevant subclasses of Monomer."""
bnagaev@378	56 for code1, is_modified, code3, name in codes:
bnagaev@378	57 cls._subclass(name, code1, code3, is_modified)
me@260	58
me@260	59 @classmethod
me@260	60 def from_code1(cls, code1):
me@328	61 """Create new monomer from 1-letter code."""
me@328	62 return cls.by_code1[code1.upper()]()
me@260	63
me@260	64 @classmethod
me@260	65 def from_code3(cls, code3):
me@328	66 """Create new monomer from 3-letter code."""
me@328	67 return cls.by_code3[code3.upper()]()
me@260	68
me@260	69 @classmethod
me@260	70 def from_name(cls, name):
me@328	71 """Create new monomer from full name."""
me@328	72 return cls.by_name[name.strip().capitalize()]()
me@260	73
me@329	74 def __repr__(self):
me@329	75 return '<Monomer %s>' % self.code3
me@329	76
me@329	77 def __str__(self):
me@329	78 """Returns one-letter code"""
me@329	79 return self.code1
me@329	80
me@260	81 def __eq__(self, other):
me@328	82 """Monomers within same monomer type are compared by code1."""
me@328	83 assert self.type == other.type
me@328	84 return self.code1 == other.code1
bnagaev@239	85
bnagaev@239	86 class Sequence(list):
me@274	87 """Sequence of Monomers.
bnagaev@243	88
me@274	89 This behaves like list of monomer objects. In addition to standard list
me@274	90 behaviour, Sequence has the following attributes:
me@270	91
me@274	92 * name -- str with the name of the sequence
me@274	93 * description -- str with description of the sequence
me@274	94 * source -- str denoting source of the sequence
me@266	95
me@274	96 Any of them may be empty (i.e. hold empty string)
me@274	97 """
me@270	98
dendik@382	99 types = base
dendik@382	100 """Mapping of related types. SHOULD be redefined in subclasses."""
me@270	101
me@275	102 name = ''
me@275	103 description = ''
me@275	104 source = ''
me@275	105
me@347	106 @classmethod
me@347	107 def from_monomers(cls, monomers=[], name=None, description=None, source=None):
me@347	108 """Create sequence from a list of monomer objecst."""
bnagaev@378	109 result = cls(monomers)
me@275	110 if name:
me@347	111 result.name = name
me@275	112 if description:
me@347	113 result.description = description
me@275	114 if source:
me@347	115 result.source = source
me@347	116 return result
me@347	117
me@347	118 @classmethod
me@347	119 def from_string(cls, string, name='', description='', source=''):
me@347	120 """Create sequences from string of one-letter codes."""
dendik@382	121 monomer = cls.types.Monomer.from_code1
me@347	122 monomers = [monomer(letter) for letter in string]
me@347	123 return cls.from_monomers(monomers, name, description, source)
me@270	124
me@329	125 def __repr__(self):
me@329	126 return '<Sequence %s>' % str(self)
me@329	127
me@262	128 def __str__(self):
me@329	129 """Returns sequence of one-letter codes."""
me@275	130 return ''.join(monomer.code1 for monomer in self)
me@270	131
me@316	132 def __hash__(self):
me@316	133 """Hash sequence by identity."""
me@316	134 return id(self)
me@316	135
me@295	136 class Alignment(object):
me@295	137 """Alignment. It is a list of Columns."""
bnagaev@249	138
dendik@382	139 types = base
dendik@382	140 """Mapping of related types. SHOULD be redefined in subclasses."""
me@288	141
me@289	142 sequences = None
me@289	143 """Ordered list of sequences in alignment. Read, but DO NOT FIDDLE!"""
bnagaev@249	144
me@287	145 def __init__(self):
me@287	146 """Initialize empty alignment."""
me@287	147 self.sequences = []
me@295	148 self.columns = []
me@282	149
me@362	150 # Alignment grow & IO methods
me@299	151 # ==============================
me@299	152
me@294	153 def append_sequence(self, sequence):
me@365	154 """Add sequence to alignment. Return self.
me@294	155
me@294	156 If sequence is too short, pad it with gaps on the right.
me@294	157 """
me@294	158 self.sequences.append(sequence)
dendik@388	159 self._pad_to_width(len(sequence))
dendik@388	160 for column, monomer in zip(self.columns, sequence):
dendik@388	161 column[sequence] = monomer
me@365	162 return self
me@294	163
me@364	164 def append_row_from_string(self, string,
me@364	165 name='', description='', source='', gaps=default_gaps):
me@364	166 """Add row from a string of one-letter codes and gaps. Return self."""
dendik@382	167 Sequence = self.types.Sequence
me@349	168 without_gaps = util.remove_each(string, gaps)
me@321	169 sequence = Sequence.from_string(without_gaps, name, description, source)
dendik@388	170 self._pad_to_width(len(string))
dendik@388	171 non_gap_columns = [column
dendik@390	172 for column, char in zip(self.columns, string)
dendik@388	173 if char not in gaps
dendik@388	174 ]
dendik@388	175 for monomer, column in zip(sequence, non_gap_columns):
dendik@388	176 column[sequence] = monomer
me@287	177 self.sequences.append(sequence)
me@364	178 return self
me@287	179
dendik@388	180 def _pad_to_width(self, n):
dendik@388	181 """Pad alignment with empty columns on the right to width n."""
dendik@388	182 for i in range(len(self.columns), n):
me@302	183 self.columns.append(Column())
me@302	184
me@362	185 def append_file(self, file, format='fasta', gaps=default_gaps):
me@365	186 """Append sequences from file to alignment. Return self.
me@299	187
me@362	188 If sequences in file have gaps (detected as characters belonging to
me@362	189 `gaps` set), treat them accordingly.
me@362	190 """
bnagaev@418	191 sequences = []
bnagaev@418	192 if format == 'fasta':
bnagaev@418	193 sequences = fileio.FastaIo(file).get_all_strings()
bnagaev@418	194 elif format == 'msf':
bnagaev@418	195 sequences = fileio.MsfIo(file).get_all_strings()
bnagaev@418	196 else:
bnagaev@418	197 raise Exception("We don't support other formats yet")
bnagaev@418	198 for (name, description, body) in sequences:
bnagaev@378	199 self.append_row_from_string(body, name, description, file.name, gaps)
me@287	200 return self
bnagaev@249	201
bnagaev@418	202 def to_file(self, file, format='fasta', gap='-'):
me@292	203 """Write alignment in FASTA file as sequences with gaps."""
me@367	204 assert format == "fasta", "We don't support other formats yet"
me@292	205 def char(monomer):
me@292	206 if monomer:
me@292	207 return monomer.code1
bnagaev@418	208 return gap
bnagaev@418	209 if format == 'fasta':
bnagaev@418	210 io = fileio.FastaIo(file)
bnagaev@418	211 elif format == 'msf':
bnagaev@418	212 io = fileio.MsfIo(file)
bnagaev@418	213 else:
bnagaev@418	214 raise Exception("We don't support other formats yet")
me@292	215 for row in self.rows_as_lists():
me@292	216 seq = row.sequence
me@292	217 line = "".join(map(char, row))
bnagaev@418	218 io.save_string(line, seq.name, seq.description)
me@292	219
me@299	220 # Data access methods for alignment
me@299	221 # =================================
me@299	222
me@299	223 def rows(self):
me@299	224 """Return list of rows (temporary objects) in alignment.
me@299	225
me@299	226 Each row is a dictionary of { column : monomer }.
me@363	227
me@299	228 For gap positions there is no key for the column in row.
me@299	229
me@299	230 Each row has attribute `sequence` pointing to the sequence the row is
me@299	231 describing.
me@299	232
me@299	233 Modifications of row have no effect on the alignment.
me@299	234 """
me@299	235 # For now, the function returns a list rather than iterator.
me@299	236 # It is yet to see, whether memory performance here becomes critical,
me@299	237 # or is random access useful.
me@299	238 rows = []
me@299	239 for sequence in self.sequences:
me@299	240 row = util.UserDict()
me@299	241 row.sequence = sequence
me@299	242 for column in self.columns:
me@299	243 if sequence in column:
me@299	244 row[column] = column[sequence]
me@299	245 rows.append(row)
me@299	246 return rows
me@299	247
me@299	248 def rows_as_lists(self):
me@299	249 """Return list of rows (temporary objects) in alignment.
me@299	250
me@299	251 Each row here is a list of either monomer or None (for gaps).
me@299	252
me@299	253 Each row has attribute `sequence` pointing to the sequence of row.
me@299	254
me@299	255 Modifications of row have no effect on the alignment.
me@299	256 """
me@299	257 rows = []
me@299	258 for sequence in self.sequences:
me@299	259 row = util.UserList()
me@299	260 row.sequence = sequence
me@299	261 for column in self.columns:
me@299	262 row.append(column.get(sequence))
me@299	263 rows.append(row)
me@299	264 return rows
me@299	265
me@299	266 def columns_as_lists(self):
me@299	267 """Return list of columns (temorary objects) in alignment.
me@299	268
me@299	269 Each column here is a list of either monomer or None (for gaps).
me@299	270
me@299	271 Items of column are sorted in the same way as alignment.sequences.
me@299	272
me@299	273 Modifications of column have no effect on the alignment.
me@299	274 """
me@299	275 columns = []
me@299	276 for column in self.columns:
me@299	277 col = []
me@299	278 for sequence in self.sequences:
me@299	279 col.append(column.get(sequence))
me@299	280 columns.append(col)
me@299	281 return columns
me@299	282
me@368	283 # Alignment / Block editing methods
me@368	284 # =================================
me@368	285
me@368	286 def _flush_row(self, row, whence='left'):
me@368	287 """Helper for `flush`: flush to one side all monomers in one row."""
me@368	288 row = filter(None, row)
me@368	289 padding = [None] * len(self.columns)
me@368	290 if whence == 'left':
me@368	291 return row + padding
me@368	292 if whence == 'right':
me@368	293 return padding + row
me@368	294 if whence == 'center':
me@368	295 pad_len = (len(self.columns) - len(row)) // 2
me@368	296 # vvv fix padding for case when length is odd: better have more
me@368	297 pad_len += len(self.columns) - 2 * pad_len
me@368	298 padding = [None] * pad_len
me@368	299 return padding + row + padding
me@368	300 assert True, "whence must be either 'left' or 'right' or 'center'"
me@368	301
me@368	302 def flush(self, whence='left'):
me@368	303 """Remove all gaps from alignment and flush results to one side.
me@368	304
me@368	305 `whence` must be one of 'left', 'right' or 'center'
me@368	306 """
me@368	307 for row in self.rows_as_lists():
me@368	308 sequence = row.sequence
me@368	309 row = self._flush_row(row, whence)
me@368	310 for monomer, column in zip(row, self.columns):
me@368	311 if monomer:
me@368	312 column[sequence] = monomer
me@368	313 elif sequence in column:
me@368	314 del column[sequence]
me@368	315
me@369	316 def remove_gap_columns(self):
me@369	317 """Remove all empty columns."""
me@369	318 for n, column in reversed(enumerate(self.columns)):
me@369	319 if column == {}:
me@369	320 self.columns[n:n+1] = []
me@369	321
me@371	322 def _wipe(self):
me@371	323 """Make all positions gaps (but keep sequences intact)."""
me@371	324 for column in self.columns:
dendik@412	325 for sequence in list(column):
me@371	326 del column[sequence]
me@371	327
me@372	328 def _merge(self, dst, new, merge):
me@373	329 """Replace contents of `dst` with those of `new`.
me@372	330
me@372	331 Replace contents of elements using function `merge(dst_el, new_le)`.
me@372	332 """
bnagaev@384	333 for el, new_el in zip(dst, new):
bnagaev@384	334 merge(el, new_el)
me@372	335 dst[len(dst):] = new[len(dst):]
me@372	336 del dst[len(new):]
me@371	337
me@373	338 def _replace_sequence_contents(self, new, copy_descriptions):
me@373	339 """Replace contents of sequences with those of `new` alignment."""
me@371	340 # XXX: we manually copy sequence contents here
me@372	341 # XXX: we only copy, overlapping parts and link to the rest
me@372	342 def merge_monomers(dst, new):
me@372	343 dst.__class__ = new.__class__
me@372	344 def merge_sequences(dst, new):
me@373	345 if copy_descriptions:
me@373	346 vars(dst).update(vars(new))
me@372	347 self._merge(dst, new, merge_monomers)
me@372	348 self._merge(self.sequences, new.sequences, merge_sequences)
me@371	349
me@371	350 def _replace_column_contents(self, new):
me@373	351 """Replace column contents with those of `new` alignment.
me@371	352
me@373	353 Synonym: copy gap patterns from `new` to `self`.
me@372	354
me@373	355 `self.sequences` and `new.sequences` should have the same contents.
me@371	356 """
me@371	357 self._wipe()
me@371	358 not_gap = lambda (a,b): a != None
me@371	359 for sequence, new_row in zip(self.sequences, new.rows_as_lists()):
me@371	360 assert len(sequence) == len(new_row.sequence)
dendik@388	361 non_gap_columns = [column
dendik@388	362 for column, monomer in zip(self.columns, new_row)
dendik@388	363 if monomer
dendik@388	364 ]
dendik@388	365 for monomer, column in zip(sequence, non_gap_columns):
dendik@388	366 column[sequence] = monomer
me@371	367
me@373	368 def _replace_contents(self, new, copy_descriptions, copy_contents):
me@371	369 """Replace alignment contents with those of other alignment."""
me@373	370 if copy_contents:
me@373	371 self._replace_sequence_contents(new, copy_descriptions)
bnagaev@378	372 self._replace_column_contents(new)
me@371	373
me@373	374 def process(self, function, copy_descriptions=True, copy_contents=True):
me@371	375 """Apply function to the alignment (or block); inject results back.
me@371	376
me@373	377 - `function(block)` must return block with same line order.
me@373	378 - if `copy_descriptions` is False, ignore new sequence names.
me@373	379 - if `copy_contents` is False, don't copy sequence contents too.
dendik@380	380
dendik@380	381 `function` (object) may have attributes `copy_descriptions` and
dendik@380	382 `copy_contents`, which override the same named arguments.
me@371	383 """
me@371	384 new = function(self)
dendik@380	385 if hasattr(function, 'copy_descriptions'):
dendik@380	386 copy_descriptions = function.copy_descriptions
dendik@380	387 if hasattr(function, 'copy_contents'):
dendik@380	388 copy_contents = function.copy_contents
me@373	389 self._replace_contents(new, copy_descriptions, copy_contents)
me@371	390
me@300	391 class Column(dict):
me@300	392 """Column of alignment.
me@300	393
me@300	394 Column is a dict of { sequence : monomer }.
me@300	395
me@300	396 For sequences that have gaps in current row, given key is not present in
me@300	397 the column.
me@300	398 """
me@325	399
dendik@382	400 types = base
dendik@382	401 """Mapping of related types. SHOULD be redefined in subclasses."""
dendik@382	402
me@325	403 def __hash__(self):
me@325	404 """Return hash by identity."""
me@325	405 return id(self)
me@300	406
me@317	407 class Block(Alignment):
me@307	408 """Block of alignment.
me@301	409
dendik@415	410 Block is an intersection of several rows & columns. (The collections of
dendik@415	411 rows and columns are represented as ordered lists, to retain display order
dendik@415	412 of Alignment or add ability to tweak it). Most of blocks look like
dendik@415	413 rectangular part of alignment if you shuffle alignment rows the right way.
me@261	414 """
me@270	415
me@307	416 alignment = None
me@307	417 """Alignment the block belongs to."""
me@270	418
me@307	419 sequences = ()
me@307	420 """List of sequences in block."""
me@307	421
me@307	422 columns = ()
me@307	423 """List of columns in block."""
me@307	424
me@317	425 @classmethod
me@317	426 def from_alignment(cls, alignment, sequences=None, columns=None):
me@307	427 """Build new block from alignment.
me@307	428
me@307	429 If sequences are not given, the block uses all sequences in alignment.
me@307	430
me@307	431 If columns are not given, the block uses all columns in alignment.
me@307	432
me@307	433 In both cases we use exactly the list used in alignment, thus, if new
me@307	434 sequences or columns are added to alignment, the block tracks this too.
me@261	435 """
me@307	436 if sequences is None:
me@307	437 sequences = alignment.sequences
me@318	438 if columns is None:
me@307	439 columns = alignment.columns
me@320	440 block = cls()
me@320	441 block.alignment = alignment
me@320	442 block.sequences = sequences
me@320	443 block.columns = columns
me@320	444 return block
me@270	445
me@260	446 # vim: set ts=4 sts=4 sw=4 et: