File size: 22,468 Bytes
b7731cd
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
# Copyright 2000 Brad Chapman.  All rights reserved.
#
# This file is part of the Biopython distribution and governed by your
# choice of the "Biopython License Agreement" or the "BSD 3-Clause License".
# Please see the LICENSE file that should have been included as part of this
# package.
"""Extract information from alignment objects.

In order to try and avoid huge alignment objects with tons of functions,
functions which return summary type information about alignments should
be put into classes in this module.
"""


import math
import sys
from collections import Counter

from Bio.Seq import Seq


class SummaryInfo:
    """Calculate summary info about the alignment.

    This class should be used to calculate information summarizing the
    results of an alignment. This may either be straight consensus info
    or more complicated things.
    """

    def __init__(self, alignment):
        """Initialize with the alignment to calculate information on.

        ic_vector attribute. A list of ic content for each column number.
        """
        self.alignment = alignment
        self.ic_vector = []

    def dumb_consensus(self, threshold=0.7, ambiguous="X", require_multiple=False):
        """Output a fast consensus sequence of the alignment.

        This doesn't do anything fancy at all. It will just go through the
        sequence residue by residue and count up the number of each type
        of residue (ie. A or G or T or C for DNA) in all sequences in the
        alignment. If the percentage of the most common residue type is
        greater then the passed threshold, then we will add that residue type,
        otherwise an ambiguous character will be added.

        This could be made a lot fancier (ie. to take a substitution matrix
        into account), but it just meant for a quick and dirty consensus.

        Arguments:
         - threshold - The threshold value that is required to add a particular
           atom.
         - ambiguous - The ambiguous character to be added when the threshold is
           not reached.
         - require_multiple - If set as True, this will require that more than
           1 sequence be part of an alignment to put it in the consensus (ie.
           not just 1 sequence and gaps).

        """
        # Iddo Friedberg, 1-JUL-2004: changed ambiguous default to "X"
        consensus = ""

        # find the length of the consensus we are creating
        con_len = self.alignment.get_alignment_length()

        # go through each seq item
        for n in range(con_len):
            # keep track of the counts of the different atoms we get
            atom_dict = Counter()
            num_atoms = 0

            for record in self.alignment:
                # make sure we haven't run past the end of any sequences
                # if they are of different lengths
                try:
                    c = record[n]
                except IndexError:
                    continue
                if c != "-" and c != ".":
                    atom_dict[c] += 1

                    num_atoms += 1

            max_atoms = []
            max_size = 0

            for atom in atom_dict:
                if atom_dict[atom] > max_size:
                    max_atoms = [atom]
                    max_size = atom_dict[atom]
                elif atom_dict[atom] == max_size:
                    max_atoms.append(atom)

            if require_multiple and num_atoms == 1:
                consensus += ambiguous
            elif len(max_atoms) == 1 and max_size / num_atoms >= threshold:
                consensus += max_atoms[0]
            else:
                consensus += ambiguous

        return Seq(consensus)

    def gap_consensus(self, threshold=0.7, ambiguous="X", require_multiple=False):
        """Output a fast consensus sequence of the alignment, allowing gaps.

        Same as dumb_consensus(), but allows gap on the output.

        Things to do:
         - Let the user define that with only one gap, the result
           character in consensus is gap.
         - Let the user select gap character, now
           it takes the same as input.

        """
        consensus = ""

        # find the length of the consensus we are creating
        con_len = self.alignment.get_alignment_length()

        # go through each seq item
        for n in range(con_len):
            # keep track of the counts of the different atoms we get
            atom_dict = Counter()
            num_atoms = 0

            for record in self.alignment:
                # make sure we haven't run past the end of any sequences
                # if they are of different lengths
                try:
                    c = record[n]
                except IndexError:
                    continue
                atom_dict[c] += 1

                num_atoms += 1

            max_atoms = []
            max_size = 0

            for atom in atom_dict:
                if atom_dict[atom] > max_size:
                    max_atoms = [atom]
                    max_size = atom_dict[atom]
                elif atom_dict[atom] == max_size:
                    max_atoms.append(atom)

            if require_multiple and num_atoms == 1:
                consensus += ambiguous
            elif len(max_atoms) == 1 and max_size / num_atoms >= threshold:
                consensus += max_atoms[0]
            else:
                consensus += ambiguous

        return Seq(consensus)

    def replacement_dictionary(self, skip_chars=None, letters=None):
        """Generate a replacement dictionary to plug into a substitution matrix.

        This should look at an alignment, and be able to generate the number
        of substitutions of different residues for each other in the
        aligned object.

        Will then return a dictionary with this information::

            {('A', 'C') : 10, ('C', 'A') : 12, ('G', 'C') : 15 ....}

        This also treats weighted sequences. The following example shows how
        we calculate the replacement dictionary. Given the following
        multiple sequence alignment::

            GTATC  0.5
            AT--C  0.8
            CTGTC  1.0

        For the first column we have::

            ('A', 'G') : 0.5 * 0.8 = 0.4
            ('C', 'G') : 0.5 * 1.0 = 0.5
            ('A', 'C') : 0.8 * 1.0 = 0.8

        We then continue this for all of the columns in the alignment, summing
        the information for each substitution in each column, until we end
        up with the replacement dictionary.

        Arguments:
         - skip_chars - Not used; setting it to anything other than None
           will raise a ValueError
         - letters - An iterable (e.g. a string or list of characters to include.
        """
        if skip_chars is not None:
            raise ValueError(
                "argument skip_chars has been deprecated; instead, please use 'letters' to specify the characters you want to include"
            )
        rep_dict = {(letter1, letter2): 0 for letter1 in letters for letter2 in letters}

        # iterate through each record
        for rec_num1 in range(len(self.alignment)):
            # iterate through each record from one beyond the current record
            # to the end of the list of records
            for rec_num2 in range(rec_num1 + 1, len(self.alignment)):
                # for each pair of records, compare the sequences and add
                # the pertinent info to the dictionary
                self._pair_replacement(
                    self.alignment[rec_num1].seq,
                    self.alignment[rec_num2].seq,
                    self.alignment[rec_num1].annotations.get("weight", 1.0),
                    self.alignment[rec_num2].annotations.get("weight", 1.0),
                    rep_dict,
                    letters,
                )

        return rep_dict

    def _pair_replacement(self, seq1, seq2, weight1, weight2, dictionary, letters):
        """Compare two sequences and generate info on the replacements seen (PRIVATE).

        Arguments:
         - seq1, seq2 - The two sequences to compare.
         - weight1, weight2 - The relative weights of seq1 and seq2.
         - dictionary - The dictionary containing the starting replacement
           info that we will modify.
         - letters - A list of characters to include when calculating replacements.

        """
        # loop through each residue in the sequences
        for residue1, residue2 in zip(seq1, seq2):
            if residue1 in letters and residue2 in letters:
                dictionary[(residue1, residue2)] += weight1 * weight2

    def _get_all_letters(self):
        """Return a string containing the expected letters in the alignment (PRIVATE)."""
        set_letters = set()
        for record in self.alignment:
            set_letters.update(record.seq)
        list_letters = sorted(set_letters)
        all_letters = "".join(list_letters)
        return all_letters

    def pos_specific_score_matrix(self, axis_seq=None, chars_to_ignore=None):
        """Create a position specific score matrix object for the alignment.

        This creates a position specific score matrix (pssm) which is an
        alternative method to look at a consensus sequence.

        Arguments:
         - chars_to_ignore - A list of all characters not to include in
           the pssm.
         - axis_seq - An optional argument specifying the sequence to
           put on the axis of the PSSM. This should be a Seq object. If nothing
           is specified, the consensus sequence, calculated with default
           parameters, will be used.

        Returns:
         - A PSSM (position specific score matrix) object.

        """
        # determine all of the letters we have to deal with
        all_letters = self._get_all_letters()
        if not all_letters:
            raise ValueError("_get_all_letters returned empty string")

        if chars_to_ignore is None:
            chars_to_ignore = []
        if not isinstance(chars_to_ignore, list):
            raise TypeError("chars_to_ignore should be a list.")

        gap_char = "-"
        chars_to_ignore.append(gap_char)

        for char in chars_to_ignore:
            all_letters = all_letters.replace(char, "")

        if axis_seq:
            left_seq = axis_seq
            if len(axis_seq) != self.alignment.get_alignment_length():
                raise ValueError(
                    "Axis sequence length does not equal the get_alignment_length"
                )
        else:
            left_seq = self.dumb_consensus()

        pssm_info = []
        # now start looping through all of the sequences and getting info
        for residue_num in range(len(left_seq)):
            score_dict = dict.fromkeys(all_letters, 0)
            for record in self.alignment:
                try:
                    this_residue = record.seq[residue_num]
                # if we hit an index error we've run out of sequence and
                # should not add new residues
                except IndexError:
                    this_residue = None

                if this_residue and this_residue not in chars_to_ignore:
                    weight = record.annotations.get("weight", 1.0)
                    try:
                        score_dict[this_residue] += weight
                    except KeyError:
                        raise ValueError(
                            "Residue %s not found" % this_residue
                        ) from None

            pssm_info.append((left_seq[residue_num], score_dict))

        return PSSM(pssm_info)

    def information_content(
        self,
        start=0,
        end=None,
        e_freq_table=None,
        log_base=2,
        chars_to_ignore=None,
        pseudo_count=0,
    ):
        """Calculate the information content for each residue along an alignment.

        Arguments:
         - start, end - The starting an ending points to calculate the
           information content. These points should be relative to the first
           sequence in the alignment, starting at zero (ie. even if the 'real'
           first position in the seq is 203 in the initial sequence, for
           the info content, we need to use zero). This defaults to the entire
           length of the first sequence.
         - e_freq_table - A dictionary specifying the expected frequencies
           for each letter (e.g. {'G' : 0.4, 'C' : 0.4, 'T' : 0.1, 'A' : 0.1}).
           Gap characters should not be included, since these should not have
           expected frequencies.
         - log_base - The base of the logarithm to use in calculating the
           information content. This defaults to 2 so the info is in bits.
         - chars_to_ignore - A listing of characters which should be ignored
           in calculating the info content. Defaults to none.

        Returns:
         - A number representing the info content for the specified region.

        Please see the Biopython manual for more information on how information
        content is calculated.

        """
        # if no end was specified, then we default to the end of the sequence
        if end is None:
            end = len(self.alignment[0].seq)
        if chars_to_ignore is None:
            chars_to_ignore = []

        if start < 0 or end > len(self.alignment[0].seq):
            raise ValueError(
                "Start (%s) and end (%s) are not in the range %s to %s"
                % (start, end, 0, len(self.alignment[0].seq))
            )
        # determine random expected frequencies, if necessary
        random_expected = None
        # determine all of the letters we have to deal with
        all_letters = self._get_all_letters()
        for char in chars_to_ignore:
            all_letters = all_letters.replace(char, "")

        info_content = {}
        for residue_num in range(start, end):
            freq_dict = self._get_letter_freqs(
                residue_num,
                self.alignment,
                all_letters,
                chars_to_ignore,
                pseudo_count,
                e_freq_table,
                random_expected,
            )
            # print(freq_dict, end="")
            column_score = self._get_column_info_content(
                freq_dict, e_freq_table, log_base, random_expected
            )
            info_content[residue_num] = column_score
        # sum up the score
        total_info = sum(info_content.values())
        # fill in the ic_vector member: holds IC for each column
        # reset ic_vector to empty list at each call
        self.ic_vector = []
        for (i, k) in enumerate(info_content):
            self.ic_vector.append(info_content[i + start])
        return total_info

    def _get_letter_freqs(
        self,
        residue_num,
        all_records,
        letters,
        to_ignore,
        pseudo_count=0,
        e_freq_table=None,
        random_expected=None,
    ):
        """Determine the frequency of specific letters in the alignment (PRIVATE).

        Arguments:
         - residue_num - The number of the column we are getting frequencies
           from.
         - all_records - All of the SeqRecords in the alignment.
         - letters - The letters we are interested in getting the frequency
           for.
         - to_ignore - Letters we are specifically supposed to ignore.
         - pseudo_count - Optional argument specifying the Pseudo count (k)
           to add in order to prevent a frequency of 0 for a letter.
         - e_freq_table - An optional argument specifying a dictionary with
           the expected frequencies for each letter.
         - random_expected - Optional argument that specify the frequency to use
           when e_freq_table is not defined.

        This will calculate the frequencies of each of the specified letters
        in the alignment at the given frequency, and return this as a
        dictionary where the keys are the letters and the values are the
        frequencies. Pseudo count can be added to prevent a null frequency
        """
        freq_info = dict.fromkeys(letters, 0)

        total_count = 0

        gap_char = "-"

        if pseudo_count < 0:
            raise ValueError(
                "Positive value required for pseudo_count, %s provided" % (pseudo_count)
            )

        # collect the count info into the dictionary for all the records
        for record in all_records:
            try:
                if record.seq[residue_num] not in to_ignore:
                    weight = record.annotations.get("weight", 1.0)
                    freq_info[record.seq[residue_num]] += weight
                    total_count += weight
            except KeyError:
                raise ValueError(
                    "Residue %s not found in letters %s"
                    % (record.seq[residue_num], letters)
                ) from None

        if e_freq_table:
            # check if all the residus in freq_info are in e_freq_table
            for key in freq_info:
                if key != gap_char and key not in e_freq_table:
                    raise ValueError("%s not found in expected frequency table" % key)

        if total_count == 0:
            # This column must be entirely ignored characters
            for letter in freq_info:
                if freq_info[letter] != 0:
                    raise ValueError("freq_info[letter] is not 0")
                # TODO - Map this to NA or NaN?
        else:
            # now convert the counts into frequencies
            for letter in freq_info:
                if pseudo_count and (random_expected or e_freq_table):
                    # use either the expected random freq or the
                    if e_freq_table:
                        ajust_freq = e_freq_table[letter]
                    else:
                        ajust_freq = random_expected

                    ajusted_letter_count = freq_info[letter] + ajust_freq * pseudo_count
                    ajusted_total = total_count + pseudo_count
                    freq_info[letter] = ajusted_letter_count / ajusted_total

                else:
                    freq_info[letter] = freq_info[letter] / total_count

        return freq_info

    def _get_column_info_content(
        self, obs_freq, e_freq_table, log_base, random_expected
    ):
        """Calculate the information content for a column (PRIVATE).

        Arguments:
         - obs_freq - The frequencies observed for each letter in the column.
         - e_freq_table - An optional argument specifying a dictionary with
           the expected frequencies for each letter.
         - log_base - The base of the logarithm to use in calculating the
           info content.

        """
        gap_char = "-"

        if e_freq_table:
            # check the expected freq information to make sure it is good
            for key in obs_freq:
                if key != gap_char and key not in e_freq_table:
                    raise ValueError(
                        f"Frequency table provided does not contain observed letter {key}"
                    )

        total_info = 0.0

        for letter in obs_freq:
            inner_log = 0.0
            # if we have expected frequencies, modify the log value by them
            # gap characters do not have expected frequencies, so they
            # should just be the observed frequency.
            if letter != gap_char:
                if e_freq_table:
                    inner_log = obs_freq[letter] / e_freq_table[letter]
                else:
                    inner_log = obs_freq[letter] / random_expected
            # if the observed frequency is zero, we don't add any info to the
            # total information content
            if inner_log > 0:
                letter_info = (
                    obs_freq[letter] * math.log(inner_log) / math.log(log_base)
                )
                total_info += letter_info
        return total_info

    def get_column(self, col):
        """Return column of alignment."""
        # TODO - Deprecate this and implement slicing?
        return self.alignment[:, col]


class PSSM:
    """Represent a position specific score matrix.

    This class is meant to make it easy to access the info within a PSSM
    and also make it easy to print out the information in a nice table.

    Let's say you had an alignment like this::

        GTATC
        AT--C
        CTGTC

    The position specific score matrix (when printed) looks like::

          G A T C
        G 1 1 0 1
        T 0 0 3 0
        A 1 1 0 0
        T 0 0 2 0
        C 0 0 0 3

    You can access a single element of the PSSM using the following::

        your_pssm[sequence_number][residue_count_name]

    For instance, to get the 'T' residue for the second element in the
    above alignment you would need to do:

    your_pssm[1]['T']
    """

    def __init__(self, pssm):
        """Initialize with pssm data to represent.

        The pssm passed should be a list with the following structure:

        list[0] - The letter of the residue being represented (for instance,
        from the example above, the first few list[0]s would be GTAT...
        list[1] - A dictionary with the letter substitutions and counts.
        """
        self.pssm = pssm

    def __getitem__(self, pos):
        return self.pssm[pos][1]

    def __str__(self):
        out = " "
        all_residues = sorted(self.pssm[0][1])

        # first print out the top header
        for res in all_residues:
            out += "   %s" % res
        out += "\n"

        # for each item, write out the substitutions
        for item in self.pssm:
            out += "%s " % item[0]
            for res in all_residues:
                out += " %.1f" % item[1][res]

            out += "\n"
        return out

    def get_residue(self, pos):
        """Return the residue letter at the specified position."""
        return self.pssm[pos][0]


def print_info_content(summary_info, fout=None, rep_record=0):
    """3 column output: position, aa in representative sequence, ic_vector value."""
    fout = fout or sys.stdout
    if not summary_info.ic_vector:
        summary_info.information_content()
    rep_sequence = summary_info.alignment[rep_record]
    for pos, (aa, ic) in enumerate(zip(rep_sequence, summary_info.ic_vector)):
        fout.write("%d %s %.3f\n" % (pos, aa, ic))