lib.py 59 KB
Newer Older
jhoogenboom's avatar
jhoogenboom committed
1
2
#!/usr/bin/env python

3
import re, sys, argparse, random
jhoogenboom's avatar
jhoogenboom committed
4
#import numpy as np  # Imported only when calling nnls()
jhoogenboom's avatar
jhoogenboom committed
5
6

from ConfigParser import RawConfigParser, MissingSectionHeaderError
7
from StringIO import StringIO
jhoogenboom's avatar
jhoogenboom committed
8
9
10
11

# Patterns that match entire sequences.
PAT_SEQ_RAW = re.compile("^[ACGT]*$")
PAT_SEQ_TSSV = re.compile("^(?:[ACGT]+\(\d+\))*$")
12
PAT_SEQ_ALLELENAME_STR = re.compile(  # First line: n_ACT[m] or alias.
jhoogenboom's avatar
jhoogenboom committed
13
    "^(?:(?:(?:CE)?-?\d+(?:\.\d+)?_(?:[ACGT]+\[\d+\])*)|((?!_).+?))"
jhoogenboom's avatar
jhoogenboom committed
14
    "(?:_[-+]\d+(?:\.1)?(?P<a>(?:(?<=\.1)-)|(?<!\.1)[ACGT]+)>"  # _+3A>
15
        "(?!(?P=a))(?:[ACGT]+|-))*$")  # Portion of variants after '>'.
16
17
18
PAT_SEQ_ALLELENAME_SNP = re.compile(
    "^REF$|^(?:(?:(?<=^)|(?<!^) )"  # 'REF' or space-separated variants.
    "\d+(?:\.1)?(?P<a>(?:(?<=\.1)-)|(?<!\.1)[ACGT]+)>"
19
        "(?!(?P=a))(?:[ACGT]+|-))+$")  # Portion of variants after '>'.
20
21
PAT_SEQ_ALLELENAME_MT = re.compile(
    "^REF$|^(?:(?:(?<=^)|(?<!^) )"  # 'REF' or space-separated variants.
22
    "(?:-?\d+\.\d+[ACGT]|(?P<a>[ACGT])?\d+(?(a)(?!(?P=a)))(?:[ACGT-]|del)))+$")
jhoogenboom's avatar
jhoogenboom committed
23

24
# Patterns that match blocks of TSSV-style sequences and allele names.
jhoogenboom's avatar
jhoogenboom committed
25
26
PAT_TSSV_BLOCK = re.compile("([ACGT]+)\((\d+)\)")
PAT_ALLELENAME_BLOCK = re.compile("([ACGT]+)\[(\d+)\]")
27
PAT_ALIAS = re.compile("^(?!_).+$")
jhoogenboom's avatar
jhoogenboom committed
28

29
30
31
32
# Patterns that match a single variant.
PAT_VARIANT_STR = re.compile(
    "^(?P<pos>[-+]\d+)(?:\.(?P<ins>1))?"
    "(?P<old>(?:(?<=\.1)-)|(?<!\.1)[ACGT]+)>"
33
    "(?!(?P=old))(?P<new>[ACGT]+|-)$")
34
35
36
PAT_VARIANT_SNP = re.compile(
    "^(?P<pos>\d+)(?:\.(?P<ins>1))?"
    "(?P<old>(?:(?<=\.1)-)|(?<!\.1)[ACGT]+)>"
37
    "(?!(?P=old))(?P<new>[ACGT]+|-)$")
38
39
40
41
PAT_VARIANT_MT = re.compile(
    "^(?P<old>(?P<a>[ACGT])|-?)"
    "(?P<pos>\d+)(?(a)|(?:\.(?P<ins>\d+))?)"
    "(?P<new>[ACGT-]|del)$")
jhoogenboom's avatar
jhoogenboom committed
42
43

# Patterns that match (parts of) an STR definition.
44
PAT_STR_DEF = re.compile("^(?:(?:(?<=^)|(?<!^)\s+)[ACGT]+\s+\d+\s+\d+)*$")
jhoogenboom's avatar
jhoogenboom committed
45
46
47
PAT_STR_DEF_BLOCK = re.compile("([ACGT]+)\s+(\d+)\s+(\d+)")

# Pattern to split a comma-, semicolon-, or space-separated list.
48
49
50
51
52
PAT_SPLIT = re.compile("\s*[,; \t]\s*")

# Pattern that matches a chromosome name/number.
PAT_CHROMOSOME = re.compile(
    "^(?:[Cc][Hh][Rr](?:[Oo][Mm])?)?([1-9XYM]|1\d|2[0-2])$")
jhoogenboom's avatar
jhoogenboom committed
53

jhoogenboom's avatar
jhoogenboom committed
54
55
# Default regular expression to capture sample tags in file names.
# This is the default of the -e command line option.
jhoogenboom's avatar
jhoogenboom committed
56
DEF_TAG_EXPR = "^(.*?)(?:\.[^.]+)?$"
jhoogenboom's avatar
jhoogenboom committed
57
58
59
60
61

# Default formatting template to write sample tags.
# This is the default of the -f command line option.
DEF_TAG_FORMAT = "\\1"

62
63
64
65
66
# Default formatting template to construct output file names for batch
# processing.  \1 and \2 refer to sample tag and tool name.
# This is the default for the -o command line option with batch support.
DEF_OUTFILE_FORMAT = "\\1-\\2.out"

jhoogenboom's avatar
jhoogenboom committed
67
68
69
70
71
72
# IUPAC Table of complementary bases.
COMPL = {"A": "T", "T": "A", "U": "A", "G": "C", "C": "G", "R": "Y", "Y": "R",
         "K": "M", "M": "K", "B": "V", "V": "B", "D": "H", "H": "D",
         "a": "t", "t": "a", "u": "a", "g": "c", "c": "g", "r": "y", "y": "r",
         "k": "m", "m": "k", "b": "v", "v": "b", "d": "h", "h": "d"}

jhoogenboom's avatar
jhoogenboom committed
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
def get_genome_pos(location, x, invert=False):
    """Get the genome position of the x-th base in a sequence."""
    if invert:
        offset = 0
        for i in range(1, len(location)):
            if i % 2:
                # Starting position.
                pos = location[i]
            elif pos <= x <= location[i]:
                # x is in the current range
                break
            else:
                offset += location[i]-pos+1
        else:
            if len(location) % 2:
                raise ValueError("Position %i is outside sequence range" % x)
        return offset + x - pos
    else:
        for i in range(1, len(location)):
            if i % 2:
                # Starting position.
                pos = location[i]
            elif location[i]-pos < x:
                # x is after this ending position
                x -= location[i]-pos+1
            else:
                # x is before this ending position
                break
        return pos + x
#get_genome_pos


106
def call_variants(template, sequence, location="suffix", cache=True,
jhoogenboom's avatar
jhoogenboom committed
107
108
109
                  debug=False):
    """
    Perform a global alignment of sequence to template and return a
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
    list of variants detected.  The format (nomenclature) of the
    returned variants depends on the location argument.

    If location is "suffix" (the default), all variants are given as
    substitutions in the form posX>Y, where the first base in the
    template is pos=1.  With location set to "prefix", bases are counted
    from right to left instead.  Insertions and deletions are written as
    pos.1->Y and posX>-, respectively.

    If location is a tuple ("M", position) with any integer for the
    position, variants are written following the mtDNA nomenclature
    guidelines.  The given position is that of the first base in the
    template.

    If location is a tuple ("chromosome name", position), a
    NotImplementedError is raised.
jhoogenboom's avatar
jhoogenboom committed
126
127
128
129
130
131
132
133
134

    By default, the results of this function are cached.  Set cache to
    False to suppress caching the result and reduce memory usage.

    Setting debug to True will cause the alignment matrices to be
    printed to sys.stdout.  Be aware that they can be quite large.
    """
    # Saving the results in a cache to avoid repeating alignments.
    try:
135
        return call_variants.cache[template, sequence, location]
jhoogenboom's avatar
jhoogenboom committed
136
    except KeyError:
137
        cache_key = location
jhoogenboom's avatar
jhoogenboom committed
138
139
140
141
142
143
144
145
146
147

    row_offset = len(template) + 1
    matrix_match = [0] * row_offset * (len(sequence)+1)
    matrix_gap1 = [-sys.maxint-1] * row_offset * (len(sequence)+1)
    matrix_gap2 = [-sys.maxint-1] * row_offset * (len(sequence)+1)

    MATCH_SCORE = 1
    MISMATCH_SCORE = -1
    GAP_OPEN_SCORE = -10
    GAP_EXTEND_SCORE = -1
148
149
150
151
152
153
154
155
    variant_format = "%i%s>%s"

    if location == "prefix":
        location = ("prefix", -len(template))
    elif location == "suffix":
        # Include plus signs for position numbers.
        variant_format = "%+i%s>%s"
        location = ("suffix", 1)
156
    elif type(location) != tuple or len(location) < 2:
157
        raise ValueError("Unknown location %r. It should be 'prefix', "
158
159
            "'suffix', or a tuple (chromosome, position [, endpos])" %
            location)
160
161
162
    elif location[0] == "M":
        # No need to avoid gaps in mtDNA notation.
        GAP_OPEN_SCORE = -1
jhoogenboom's avatar
jhoogenboom committed
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

    for i in range(len(matrix_match)):
        x = i % row_offset
        y = i / row_offset

        # Initialisation of first row and column.
        if x == 0 or y == 0:
            if x != 0:
                # Top row.
                matrix_gap1[i] = GAP_OPEN_SCORE + GAP_EXTEND_SCORE * (x - 1)
                matrix_match[i] = matrix_gap1[i]
            elif y != 0:
                # Left column.
                matrix_gap2[i] = GAP_OPEN_SCORE + GAP_EXTEND_SCORE * (y - 1)
                matrix_match[i] = matrix_gap2[i]
            continue

        matrix_gap1[i] = max(
            matrix_match[i-1] + GAP_OPEN_SCORE,
            matrix_gap1[i-1] + GAP_EXTEND_SCORE)
        matrix_gap2[i] = max(
            matrix_match[i-row_offset] + GAP_OPEN_SCORE,
            matrix_gap2[i-row_offset] + GAP_EXTEND_SCORE)

        if template[x-1] == sequence[y-1]:
            match = MATCH_SCORE
        else:
            match = MISMATCH_SCORE

        matrix_match[i] = max(
            matrix_match[i-1-row_offset] + match,
            matrix_gap1[i],
            matrix_gap2[i])

    if debug:
        print("GAP1")
        for i in range(0, len(matrix_gap1), row_offset):
            print(("%5i" * row_offset) % tuple(matrix_gap1[i:i+row_offset]))
        print("GAP2")
        for i in range(0, len(matrix_gap2), row_offset):
            print(("%5i" * row_offset) % tuple(matrix_gap2[i:i+row_offset]))
        print("Match")
        for i in range(0, len(matrix_match), row_offset):
            print(("%5i" * row_offset) % tuple(matrix_match[i:i+row_offset]))


    # Backtracking.
    variants = []
    variant_template = 0
    variant_sequence = 0
    i = len(matrix_match) - 1
    while i >= 0:
        x = i % row_offset
        y = i / row_offset

        if matrix_gap1[i] == matrix_match[i]:
            # Go horizontally.  Deletion.
            variant_template += 1
            i -= 1
            continue

        if matrix_gap2[i] == matrix_match[i]:
            # Go vertically.  Insertion.
            variant_sequence += 1
            i -= row_offset
            continue

        # Only backtracking diagonally if a gap is out of the question.
        # Go diagonally.  Either match or mismatch.
        if i == 0 or template[x - 1] == sequence[y - 1]:
            # Match.  Flush variants.
            if variant_template or variant_sequence:
235
236
237
238
239
240
                if location[0] == "M":
                    # MtDNA variants are one-base-at-a-time.
                    for j in range(
                            max(variant_template, variant_sequence)-1, -1, -1):
                        variants.append("%s%i%s%s" % (
                            template[x+j] if j < variant_template else "",#"-",
241
242
                            get_genome_pos(
                                location, x + min(j, variant_template-1)),
243
244
                            ".%i" % (j-variant_template+1)
                                if j >= variant_template else "",
245
                            sequence[y+j] if j < variant_sequence else "del"))
246
                elif variant_template == 0:
jhoogenboom's avatar
jhoogenboom committed
247
                    # Insertions: "-131.1->C" instead of "-130->C".
248
                    variants.append(variant_format % (
249
                        get_genome_pos(location, x - 1),
250
                        ".1-",
jhoogenboom's avatar
jhoogenboom committed
251
252
                        sequence[y:y+variant_sequence]))
                else:
253
                    variants.append(variant_format % (
254
                        get_genome_pos(location, x),
jhoogenboom's avatar
jhoogenboom committed
255
256
257
258
259
260
261
262
263
264
                        template[x:x+variant_template],
                        sequence[y:y+variant_sequence] or "-"))
                variant_template = 0
                variant_sequence = 0
        else:
            # Start/extend mismatch.
            variant_template += 1
            variant_sequence += 1
        i -= 1 + row_offset

265
266
    # Variants were called from right to left.  Reverse their order.
    if location[0] != "prefix":
jhoogenboom's avatar
jhoogenboom committed
267
268
269
270
        variants.reverse()

    # Store the result in the cache.
    if cache:
271
        call_variants.cache[template, sequence, cache_key] = variants
jhoogenboom's avatar
jhoogenboom committed
272
273
274
275
276
    return variants
#call_variants
call_variants.cache = {}


277
def mutate_sequence(seq, variants, location=None):
jhoogenboom's avatar
jhoogenboom committed
278
    """Apply the given variants to the given sequence."""
279
    if type(location) != tuple or len(location) < 2:
280
        pattern = PAT_VARIANT_STR
281
        location = (None, 0)
282
283
    elif location[0] == "M":
        pattern = PAT_VARIANT_MT
284
        location = (location[0], location[1]-1) + tuple(location[2:])
285
286
    else:
        pattern = PAT_VARIANT_SNP
287
        location = (location[0], location[1]-1) + tuple(location[2:])
288
289

    seq = [[]] + [[base] for base in seq]
jhoogenboom's avatar
jhoogenboom committed
290
    for variant in variants:
291
        vm = pattern.match(variant)
jhoogenboom's avatar
jhoogenboom committed
292
293
        if vm is None:
            raise ValueError("Unrecognised variant '%s'" % variant)
294
295
296
297
        pos = int(vm.group("pos"))
        ins = int(vm.group("ins") or 0)
        old = vm.group("old")
        new = vm.group("new")
jhoogenboom's avatar
jhoogenboom committed
298
299
        if old == "-":
            old = ""
300
        if new == "-" or new == "del":
jhoogenboom's avatar
jhoogenboom committed
301
302
            new = ""
        if pos < 0:
303
            pos += len(seq) + 1
304
305
        pos = get_genome_pos(location, pos, True)
        if pos < 0 or (pos == 0 and not ins) or pos >= len(seq):
306
            raise ValueError(
307
308
                "Position of variant '%s' is outside sequence range" %
                    (variant))
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
        if (not ins and old and old != "".join("".join(x[:1])
                for x in seq[pos:pos+len(old)])):
            raise ValueError(
                "Incorrect original sequence in variant '%s'; should be '%s'!"
                % (variant, "".join("".join(x[:1])
                    for x in seq[pos:pos+len(old)])))
        elif not ins and not old:
            # MtDNA substitution with reference base omitted.
            old = "".join("".join(x[:1]) for x in seq[pos:pos+len(new)])
        if not ins:
            # Remove old bases, retaining those inserted between/after.
            seq[pos:pos+len(old)] = [
                [""] + x[1:] for x in seq[pos:pos+len(old)]]
            # Place new entirely in the position of the first old base.
            seq[pos][0] = new
jhoogenboom's avatar
jhoogenboom committed
324
        else:
325
326
327
328
329
            # Insert new exactly ins positions after pos.
            while len(seq[pos]) <= ins:
                seq[pos].append("")
            seq[pos][ins] = new
    return "".join("".join(x) for x in seq)
jhoogenboom's avatar
jhoogenboom committed
330
331
332
#mutate_sequence


333
def parse_library(libfile, stream=False):
jhoogenboom's avatar
jhoogenboom committed
334
    try:
335
        if not stream:
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
            libfile = sys.stdin if libfile == "-" else open(libfile, "r")
        if libfile == sys.stdin:
            # Can't seek on pipes, so read it into a buffer first.
            libfile = StringIO(sys.stdin.read())
        try:
            library = parse_library_ini(libfile)
            if not stream:
                libfile.close()
            return library
        except MissingSectionHeaderError:
            # Not an ini file.
            pass
        libfile.seek(0)
        library = parse_library_tsv(libfile)
        if not stream and libfile != sys.stdin:
351
352
            libfile.close()
        return library
353
354
    except ValueError as err:
        raise argparse.ArgumentTypeError(err)
jhoogenboom's avatar
jhoogenboom committed
355
356
357
358
359
360
361
362
363
364
365
366
367
368
#parse_library


def parse_library_tsv(handle):
    """
    Parse a TSSV library file (tab-separated values format).

    The provided file should contain at least four columns: marker name,
    left flanking sequence, right flanking sequence, and STR definition.

    Return a nested dict with top-level keys "flanks" and "regex".
    """
    library = {
      "flanks": {},
369
370
      "regex": {},
      "regex_middle": {}
jhoogenboom's avatar
jhoogenboom committed
371
372
    }
    for line in handle:
jhoogenboom's avatar
jhoogenboom committed
373
        line = [x.strip() for x in line.rstrip("\r\n").split("\t")]
jhoogenboom's avatar
jhoogenboom committed
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
        if line == [""]:
            continue
        if len(line) < 4:
            raise ValueError(
                "Invalid library file: encountered line with %i columns, "
                "need at least 4" % len(line))
        marker = line[0]
        if PAT_SEQ_RAW.match(line[1]) is None:
            raise ValueError("Flanking sequence '%s' of marker %s is invalid" %
                             (line[1], marker))
        if PAT_SEQ_RAW.match(line[2]) is None:
            raise ValueError("Flanking sequence '%s' of marker %s is invalid" %
                             (line[2], marker))
        if PAT_STR_DEF.match(line[3]) is None:
            raise ValueError("STR definition '%s' of marker %s is invalid" %
                             (line[3], marker))
        library["flanks"][marker] = line[1:3]
jhoogenboom's avatar
jhoogenboom committed
391
392
        library["regex_middle"][marker] = re.compile("".join(
            "(%s){%s,%s}" % x for x in PAT_STR_DEF_BLOCK.findall(line[3])))
393
394
        library["regex"][marker] = re.compile(
            "".join(["^", library["regex_middle"][marker].pattern, "$"]))
jhoogenboom's avatar
jhoogenboom committed
395
396
397
398
399
400
401
402
403
404
    return library
#parse_library_tsv


def parse_library_ini(handle):
    library = {
      "flanks": {},
      "prefix": {},
      "suffix": {},
      "regex": {},
405
      "regex_middle": {},
406
407
      "nostr_reference": {},
      "genome_position": {},
408
409
      "length_adjust": {},
      "block_length": {},
410
      "max_expected_copies": {},
jhoogenboom's avatar
jhoogenboom committed
411
412
413
414
415
416
417
418
419
420
      "aliases": {}
    }
    markers = set()

    ini = RawConfigParser()
    ini.optionxform = str
    ini.readfp(handle)
    for section in ini.sections():
        for marker in ini.options(section):
            value = ini.get(section, marker)
421
422
            section_low = section.lower()
            if section_low == "flanks":
jhoogenboom's avatar
jhoogenboom committed
423
424
425
426
427
428
429
430
431
432
433
434
                values = PAT_SPLIT.split(value)
                if len(values) != 2:
                    raise ValueError(
                        "For marker %s, %i flanking sequences were given,"
                        "need exactly 2" % (marker, len(values)))
                for value in values:
                    if PAT_SEQ_RAW.match(value) is None:
                        raise ValueError(
                            "Flanking sequence '%s' of marker %s is invalid" %
                            (value, marker))
                library["flanks"][marker] = values
                markers.add(marker)
435
            elif section_low == "prefix":
436
437
438
                if marker in library["nostr_reference"]:
                    raise ValueError(
                        "A prefix was defined for non-STR marker %s" % marker)
jhoogenboom's avatar
jhoogenboom committed
439
440
441
442
443
444
                values = PAT_SPLIT.split(value)
                for value in values:
                    if PAT_SEQ_RAW.match(value) is None:
                        raise ValueError(
                            "Prefix sequence '%s' of marker %s is invalid" %
                            (value, marker))
445
                library["prefix"][marker] = values
jhoogenboom's avatar
jhoogenboom committed
446
                markers.add(marker)
447
            elif section_low == "suffix":
448
449
450
                if marker in library["nostr_reference"]:
                    raise ValueError(
                        "A suffix was defined for non-STR marker %s" % marker)
jhoogenboom's avatar
jhoogenboom committed
451
452
453
454
455
456
                values = PAT_SPLIT.split(value)
                for value in values:
                    if PAT_SEQ_RAW.match(value) is None:
                        raise ValueError(
                            "Suffix sequence '%s' of marker %s is invalid" %
                            (value, marker))
457
                library["suffix"][marker] = values
jhoogenboom's avatar
jhoogenboom committed
458
                markers.add(marker)
459
460
            elif section_low == "genome_position":
                values = PAT_SPLIT.split(value)
461
                if len(values) < 2:
462
463
                    raise ValueError(
                        "Invalid genome position '%s' for marker %s. Expected "
464
                        "a chromosome name and one or more position numbers." %
465
466
467
468
469
470
                        (value, marker))
                chromosome = PAT_CHROMOSOME.match(values[0])
                if chromosome is None:
                    raise ValueError(
                        "Invalid chromosome '%s' for marker %s." %
                        (values[0], marker))
471
472
473
474
475
476
477
478
479
480
481
482
483
484
                pos = [chromosome.group(1)]
                for i in range(1, len(values)):
                    try:
                        pos.append(int(values[i]))
                    except:
                        raise ValueError(
                            "Position '%s' of marker %s is not a valid integer"
                            % (values[i], marker))
                    if not i % 2 and pos[-2] >= pos[-1]:
                        raise ValueError(
                            "End position %i of marker %s must be higher than "
                            "corresponding start position %i" %
                            (pos[-1], marker, pos[-2]))
                library["genome_position"][marker] = tuple(pos)
485
                markers.add(marker)
486
            elif section_low == "length_adjust":
jhoogenboom's avatar
jhoogenboom committed
487
488
489
490
491
492
493
494
                try:
                    value = int(value)
                except:
                    raise ValueError(
                        "Length adjustment '%s' of marker %s is not a valid "
                        "integer" % (value, marker))
                library["length_adjust"][marker] = value
                markers.add(marker)
495
            elif section_low == "block_length":
jhoogenboom's avatar
jhoogenboom committed
496
497
498
499
500
501
502
503
                try:
                    value = int(value)
                except:
                    raise ValueError(
                        "Block length '%s' of marker %s is not a valid integer"
                        % (value, marker))
                library["block_length"][marker] = value
                markers.add(marker)
504
505
506
507
508
509
510
511
512
513
            elif section_low == "max_expected_copies":
                try:
                    value = int(value)
                except:
                    raise ValueError(
                        "Maximum number of expected copies '%s' of marker %s "
                        "is not a valid integer" % (value, marker))
                library["max_expected_copies"][marker] = value
                markers.add(marker)
            elif section_low == "aliases":
jhoogenboom's avatar
jhoogenboom committed
514
515
516
517
518
                values = PAT_SPLIT.split(value)
                if len(values) != 3:
                    raise ValueError("Alias %s does not have 3 values, but %i"
                                     % (marker, len(values)))
                if PAT_SEQ_RAW.match(values[1]) is None:
519
520
521
522
523
524
525
                    raise ValueError(
                        "Alias sequence '%s' of alias %s is invalid" %
                        (values[1], marker))
                if PAT_ALIAS.match(values[2]) is None:
                    raise ValueError(
                        "Allele name '%s' of alias %s is invalid" %
                        (values[2], marker))
jhoogenboom's avatar
jhoogenboom committed
526
527
528
529
530
531
                library["aliases"][marker] = {
                    "marker": values[0],
                    "sequence": values[1],
                    "name": values[2]
                }
                markers.add(marker)
532
            elif section_low == "repeat":
533
534
535
536
                if marker in library["nostr_reference"]:
                    raise ValueError(
                        "Marker %s was encountered in both [repeat] and "
                        "[no_repeat] sections" % marker)
jhoogenboom's avatar
jhoogenboom committed
537
538
539
540
541
542
                if PAT_STR_DEF.match(value) is None:
                    raise ValueError(
                        "STR definition '%s' of marker %s is invalid" %
                        (value, marker))
                library["regex"][marker] = value
                markers.add(marker)
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
            elif section_low == "no_repeat":
                if marker in library["regex"]:
                    raise ValueError(
                        "Marker %s was encountered in both [repeat] and "
                        "[no_repeat] sections" % marker)
                if marker in library["prefix"] or marker in library["suffix"]:
                    raise ValueError(
                        "A prefix or suffix was defined for non-STR marker %s"
                        % marker)
                if PAT_SEQ_RAW.match(value) is None:
                    raise ValueError(
                        "Reference sequence '%s' of marker %s is invalid" %
                        (value, marker))
                library["nostr_reference"][marker] = value
                markers.add(marker)

    # Sanity check: prohibit prefix/suffix for aliases of non-STRs.
    for alias in library["aliases"]:
        if library["aliases"][alias]["marker"] in library["nostr_reference"] \
                and (alias in library["prefix"] or alias in library["suffix"]):
            raise ValueError(
                "A prefix or suffix was defined for alias %s of non-STR "
                "marker %s" % (alias, library["aliases"][alias]["marker"]))
jhoogenboom's avatar
jhoogenboom committed
566

567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
    # Sanity check: end position of marker should reflect ref length.
    for marker in library["genome_position"]:
        if marker not in library["nostr_reference"]:
            continue
        pos = library["genome_position"][marker]
        reflength = len(library["nostr_reference"][marker])
        length = 0
        for i in range(2, len(pos), 2):
            length += pos[i] - pos[i-1] + 1
        if reflength < length or (len(pos) % 2 and reflength != length):
            raise ValueError(
                "Length of reference sequence of marker %s is %i bases, but "
                "genome positions add up to %i bases" %
                (marker, reflength, length))

jhoogenboom's avatar
jhoogenboom committed
582
583
584
585
586
587
    # Compile regular expressions.
    # NOTE: The libconvert tool expects "(seq){num,num}" blocks ONLY!
    # TODO: Should a single prefix/suffix be required (i.e., seq{1,1})?
    #       Then also update libconvert when converting to TSSV format.
    for marker in markers:
        parts = []
588
        partsm = []
jhoogenboom's avatar
jhoogenboom committed
589
        if marker in library["prefix"]:
jhoogenboom's avatar
jhoogenboom committed
590
            parts += ("(%s){0,1}" % x for x in library["prefix"][marker])
jhoogenboom's avatar
jhoogenboom committed
591
592
        if marker in library["aliases"]:
            parts.append("(%s){0,1}" % library["aliases"][marker]["sequence"])
593
            partsm.append("(%s){0,1}" % library["aliases"][marker]["sequence"])
jhoogenboom's avatar
jhoogenboom committed
594
        elif marker in library["regex"]:
595
596
            partsm = ["(%s){%s,%s}" % x for x in
                      PAT_STR_DEF_BLOCK.findall(library["regex"][marker])]
597
            parts += partsm
jhoogenboom's avatar
jhoogenboom committed
598
        if marker in library["suffix"]:
jhoogenboom's avatar
jhoogenboom committed
599
            parts += ("(%s){0,1}" % x for x in library["suffix"][marker])
jhoogenboom's avatar
jhoogenboom committed
600
601
602
        if parts:
            library["regex"][marker] = re.compile(
                "".join(["^"] + parts + ["$"]))
603
604
        if partsm:
            library["regex_middle"][marker] = re.compile("".join(partsm))
jhoogenboom's avatar
jhoogenboom committed
605
606
    return library
#parse_library_ini
jhoogenboom's avatar
jhoogenboom committed
607
608


609
def load_profiles(profilefile, library=None):
jhoogenboom's avatar
jhoogenboom committed
610
    column_names = profilefile.readline().rstrip("\r\n").split("\t")
611
612
613
    (colid_marker, colid_allele, colid_sequence, colid_fmean, colid_rmean,
     colid_tool) = get_column_ids(column_names, "marker", "allele", "sequence",
        "fmean", "rmean", "tool")
jhoogenboom's avatar
jhoogenboom committed
614
615
616
617
618
619
620
621
622
623
624
625
626

    profiles = {}
    for line in profilefile:
        line = line.rstrip("\r\n").split("\t")
        if line == [""]:
            continue
        marker = line[colid_marker]
        if marker not in profiles:
            profiles[marker] = {
                "m": set(),  # To be replaced by its length below.
                "n": set(),  # To be replaced by its length below.
                "seqs": [],
                "forward": {},  # To be replaced by a list below.
627
628
                "reverse": {},  # To be replaced by a list below.
                "tool": {}  # To be replaced by a list below.
jhoogenboom's avatar
jhoogenboom committed
629
630
631
632
633
634
635
636
637
638
639
640
                }
        allele = ensure_sequence_format(line[colid_allele], "raw",
            library=library, marker=marker)
        sequence = ensure_sequence_format(line[colid_sequence], "raw",
            library=library, marker=marker)
        if (allele, sequence) in profiles[marker]["forward"]:
            raise ValueError(
                "Invalid background noise profiles file: encountered "
                "multiple values for marker '%s' allele '%s' sequence '%s'" %
                (marker, allele, sequence))
        profiles[marker]["forward"][allele,sequence] = float(line[colid_fmean])
        profiles[marker]["reverse"][allele,sequence] = float(line[colid_rmean])
641
        profiles[marker]["tool"][allele, sequence] = line[colid_tool]
jhoogenboom's avatar
jhoogenboom committed
642
643
644
645
646
647
648
649
650
651
        profiles[marker]["m"].update((allele, sequence))
        profiles[marker]["n"].add(allele)

    # Check completeness and reorder true alleles.
    for marker in profiles:
        profiles[marker]["seqs"] = list(profiles[marker]["n"]) + \
            list(profiles[marker]["m"]-profiles[marker]["n"])
        profiles[marker]["n"] = len(profiles[marker]["n"])
        profiles[marker]["m"] = len(profiles[marker]["m"])
        newprofiles = {"forward": [], "reverse": []}
652
        tools = []
jhoogenboom's avatar
jhoogenboom committed
653
654
655
656
        for i in range(profiles[marker]["n"]):
            allele = profiles[marker]["seqs"][i]
            for direction in newprofiles:
                newprofiles[direction].append([0] * profiles[marker]["m"])
657
            tools.append([""] * profiles[marker]["m"])
jhoogenboom's avatar
jhoogenboom committed
658
659
660
661
662
663
            for j in range(profiles[marker]["m"]):
                sequence = profiles[marker]["seqs"][j]
                if (allele, sequence) in profiles[marker]["forward"]:
                    for direction in newprofiles:
                        newprofiles[direction][i][j] = \
                            profiles[marker][direction][allele, sequence]
664
                    tools[i][j] = profiles[marker]["tool"][allele, sequence]
jhoogenboom's avatar
jhoogenboom committed
665
666
        profiles[marker]["forward"] = newprofiles["forward"]
        profiles[marker]["reverse"] = newprofiles["reverse"]
667
        profiles[marker]["tool"] = tools
jhoogenboom's avatar
jhoogenboom committed
668
669

    return profiles
670
#load_profiles
671
672


673
674
675
676
677
678
679
680
def regex_longest_match(pattern, subject):
    """Return the longest match of the pattern in the subject string."""
    return reduce(lambda x, y:
        y if x is None or x.end()-x.start() < y.end()-y.start() else x,
        pattern.finditer(subject), None)
#regex_longest_match


jhoogenboom's avatar
jhoogenboom committed
681
682
683
def detect_sequence_format(seq):
    """Return format of seq.  One of 'raw', 'tssv', or 'allelename'."""
    if not seq:
jhoogenboom's avatar
jhoogenboom committed
684
        raise ValueError("Empty sequence")
jhoogenboom's avatar
jhoogenboom committed
685
686
687
688
    if PAT_SEQ_RAW.match(seq):
        return 'raw'
    if PAT_SEQ_TSSV.match(seq):
        return 'tssv'
689
690
    if PAT_SEQ_ALLELENAME_STR.match(seq) or PAT_SEQ_ALLELENAME_MT.match(seq) \
            or PAT_SEQ_ALLELENAME_SNP.match(seq):
jhoogenboom's avatar
jhoogenboom committed
691
        return 'allelename'
jhoogenboom's avatar
jhoogenboom committed
692
    raise ValueError("Unrecognised sequence format")
jhoogenboom's avatar
jhoogenboom committed
693
694
695
#detect_sequence_format


jhoogenboom's avatar
jhoogenboom committed
696
def convert_sequence_tssv_raw(seq):
jhoogenboom's avatar
jhoogenboom committed
697
698
    return "".join(block[0] * int(block[1])
                   for block in PAT_TSSV_BLOCK.findall(seq))
jhoogenboom's avatar
jhoogenboom committed
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
#convert_sequence_tssv_raw


def convert_sequence_raw_tssv(seq, library, marker, return_alias=False):
    # Try to match this marker's pattern, or any of its aliases.
    match = None
    if "aliases" in library:
        for alias in library["aliases"]:
            if (library["aliases"][alias]["marker"] == marker and
                    alias in library["regex"]):
                match = library["regex"][alias].match(seq)
                if match is not None:
                    marker = alias
                    break
    if match is None and marker in library["regex"]:
        match = library["regex"][marker].match(seq)
715
    if match is not None:
jhoogenboom's avatar
jhoogenboom committed
716
717
        parts = ((match.group(i), match.end(i)) for i in range(1, 1 if
            match.lastindex is None else match.lastindex+1) if match.group(i))
jhoogenboom's avatar
jhoogenboom committed
718

719
    # Use heuristics if the sequence does not match the pattern.
jhoogenboom's avatar
jhoogenboom committed
720
    else:
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741

        # Find explictily provided prefix and/or suffix if present.
        pre_suf = ["", ""]
        if "prefix" in library and marker in library["prefix"]:
            for prefix in library["prefix"][marker]:
                if seq.startswith(prefix):
                    pre_suf[0] = prefix
                    seq = seq[len(prefix):]
                    break
        if "suffix" in library and marker in library["suffix"]:
            for suffix in library["suffix"][marker]:
                if seq.endswith(suffix):
                    pre_suf[1] = suffix
                    seq = seq[:-len(suffix)]
                    break

        # Find longest match of middle pattern.
        middle = [(seq, len(pre_suf[0])+len(seq))]
        if marker in library["regex_middle"]:
            match = regex_longest_match(library["regex_middle"][marker], seq)
            if match is not None and match.end()-match.start():
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810

                # If this allele does not match the prefix of this
                # marker, but the canonical prefix of the marker ends
                # with the same sequence as the start of our match, we
                # move that portion of the match into the prefix.
                # Then, we do the same thing with the suffix.
                matched = match.group()
                start = match.start()
                end = match.end()
                modified = False
                if (not pre_suf[0] and "prefix" in library
                        and marker in library["prefix"]):
                    ref = library["prefix"][marker][0]
                    i = min(len(ref), len(matched))
                    while i > 0:
                        if ref.endswith(matched[:i]):
                            start += i
                            matched = matched[i:]
                            modified = True
                            break
                        i -= 1
                if (not pre_suf[1] and "suffix" in library
                        and marker in library["suffix"]):
                    ref = library["suffix"][marker][0]
                    i = min(len(ref), len(matched))
                    while i > 0:
                        if ref.startswith(matched[-i:]):
                            end -= i
                            matched = matched[:-i]
                            modified = True
                            break
                        i -= 1
                if modified:
                    from_start = start-match.start()
                    from_end = match.end()-end
                    middle = reduce(
                        lambda x, i: (
                            x[0] + [match.group(i)] *
                                ((match.end(i)-x[1])/len(match.group(i))),
                            match.end(i)) if match.group(i) else x,
                        range(1, match.lastindex+1), ([], match.start()))[0]
                    while from_start:
                        if from_start < len(middle[0]):
                            middle[0] = middle[0][from_start:]
                            break
                        else:
                            from_start -= len(middle[0])
                            middle = middle[1:]
                    while from_end:
                        if from_end < len(middle[-1]):
                            middle[-1] = middle[-1][:-from_end]
                            break
                        else:
                            from_end -= len(middle[-1])
                            middle = middle[:-1]
                    if middle:
                        middle = reduce(
                            lambda x, y: (x[:-1] if x[-1][0] == y else x) +
                                [(y, x[-1][1]+len(y))], middle[1:],
                                [(middle[0], start+len(middle[0]))])

                else:
                    # No trickery with prefix or suffix was done.
                    middle = [(match.group(i), match.end(i)+len(pre_suf[0]))
                        for i in range(1, match.lastindex+1) if match.group(i)]

                pre_suf[0] += seq[:start]
                pre_suf[1] = seq[end:] + pre_suf[1]
                seq = matched
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825

        # Now construct parts.
        parts = []
        if pre_suf[0]:
            parts.append((pre_suf[0], len(pre_suf[0])))
        parts += middle
        if pre_suf[1]:
            parts.append((pre_suf[1], sum(map(len,pre_suf))+len(seq)))

    seq = reduce(
        lambda a, b: (a[0] + "%s(%i)" % (b[0], (b[1]-a[1])/len(b[0])), b[1]),
        reduce(
            lambda x, y: x[:-1] + [y] if x[-1][0] == y[0] else x + [y],
            parts,
            [("", 0)]))[0]
jhoogenboom's avatar
jhoogenboom committed
826
827
828
829
830
831
    return (seq, marker) if return_alias else seq
#convert_sequence_raw_tssv


def convert_sequence_allelename_tssv(seq, library, marker):
    # Check whether there is an alias for this sequence.
832
    alias_of = None
jhoogenboom's avatar
jhoogenboom committed
833
834
835
836
837
    if "aliases" in library:
        for alias in library["aliases"]:
            if library["aliases"][alias]["marker"] == marker and (
                    seq == library["aliases"][alias]["name"] or
                    seq.startswith(library["aliases"][alias]["name"] + "_")):
838
                alias_of = marker
jhoogenboom's avatar
jhoogenboom committed
839
840
841
842
843
844
845
                marker = alias
                seq = "".join([
                    "0_",
                    library["aliases"][alias]["sequence"] + "[1]",
                    seq[len(library["aliases"][alias]["name"]):]])
                break

846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
    nameformat = None
    if PAT_SEQ_ALLELENAME_MT.match(seq) is not None:
        nameformat = "MtDNA"
    elif PAT_SEQ_ALLELENAME_SNP.match(seq) is not None:
        nameformat = "SNP"
    if nameformat is not None:
        # MtDNA and SNP markers.
        try:
            reference = library["nostr_reference"][marker]
        except KeyError:
            raise ValueError(
                "%s allele '%s' found for marker %s, but "
                "no reference sequence was found in the library" %
                (nameformat, seq, marker))
        if seq == "REF":
            return reference + "(1)"
        return mutate_sequence(reference, seq.split(),
            library["genome_position"].get(marker,
                ("M" if nameformat == "MtDNA" else "", 1))) + "(1)"

    # Note: aliases of mtDNA and SNP markers end up here as well.
    # It should NOT look like an alias now, however.
    match = PAT_SEQ_ALLELENAME_STR.match(seq)
869
870
871
    if match is None or match.group(1) is not None:
        raise ValueError("Invalid allele name '%s' encountered!" % seq)

jhoogenboom's avatar
jhoogenboom committed
872
873
874
    allele = seq.split("_")

    # Get and mutate prefix and suffix.
875
876
877
878
879
880
881
882
883
884
885
886
    prefix = ""
    suffix = ""
    if "prefix" in library:
        if marker in library["prefix"]:
            prefix = library["prefix"][marker][0]
        elif alias_of is not None and alias_of in library["prefix"]:
            prefix = library["prefix"][alias_of][0]
    if "suffix" in library:
        if marker in library["suffix"]:
            suffix = library["suffix"][marker][0]
        elif alias_of is not None and alias_of in library["suffix"]:
            suffix = library["suffix"][alias_of][0]
jhoogenboom's avatar
jhoogenboom committed
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
    variants = [[], []]
    for variant in allele[2:]:
        if variant[0] == "-":
            if not prefix:
                raise ValueError("Encountered prefix variant '%s', but marker "
                                 "'%s' has no prefix!" % (variant, marker))
            variants[0].append(variant)
        elif variant[0] == "+":
            if not suffix:
                raise ValueError("Encountered suffix variant '%s', but marker "
                                 "'%s' has no suffix!" % (variant, marker))
            variants[1].append(variant)
        else:
            raise ValueError("Unrecognised variant '%s'" % variant)
    if variants[0]:
        prefix = mutate_sequence(prefix, variants[0])
    if variants[1]:
        suffix = mutate_sequence(suffix, variants[1])

    blocks = []
    if prefix:
        blocks.append((prefix, 1))
    for block in PAT_ALLELENAME_BLOCK.findall(allele[1]):
        blocks.append((block[0], int(block[1])))
    if suffix:
        blocks.append((suffix, 1))
jhoogenboom's avatar
jhoogenboom committed
913
    return "".join("%s(%i)" % block for block in blocks)
jhoogenboom's avatar
jhoogenboom committed
914
915
916
917
918
919
#convert_sequence_allelename_tssv


def convert_sequence_raw_allelename(seq, library, marker):
    # We actually convert raw->allelename via TSSV format.
    seq, alias = convert_sequence_raw_tssv(seq, library, marker, True)
920
921
    blocks = PAT_TSSV_BLOCK.findall(seq)

922
923
924
925
    if marker in library["nostr_reference"]:
        # Handle non-STR markers here.
        if alias != marker:
            return library["aliases"][alias]["name"]
926
927
928
        if not blocks:
            # Oh dear, empty sequence... Primer dimer?
            blocks = (("",),)
929
930
931
932
933
934
        if library["nostr_reference"][marker] == blocks[0][0]:
            return "REF"
        return " ".join(
            call_variants(library["nostr_reference"][marker], blocks[0][0],
                library["genome_position"].get(marker, "suffix")))

935
936
937
    # Find prefix and suffix.
    prefix = suffix = this_prefix = this_suffix = ""
    remaining_blocks = len(blocks)
938
939
940
941
942
    if "prefix" in library:
        if alias in library["prefix"]:
            prefix = library["prefix"][alias][0]
        elif marker in library["prefix"]:
            prefix = library["prefix"][marker][0]
jhoogenboom's avatar
jhoogenboom committed
943
        if prefix and remaining_blocks > 0 and blocks[0][1] == "1":
944
            remaining_blocks -= 1
945
946
947
948
949
    if "suffix" in library:
        if alias in library["suffix"]:
            suffix = library["suffix"][alias][0]
        elif marker in library["suffix"]:
            suffix = library["suffix"][marker][0]
jhoogenboom's avatar
jhoogenboom committed
950
        if suffix and remaining_blocks > 0 and blocks[-1][1] == "1":
951
            remaining_blocks -= 1
952
953
954
955
956
957
    if remaining_blocks > 0 and prefix and blocks[0][1] == "1":
        this_prefix = blocks[0][0]
        blocks = blocks[1:]
    if remaining_blocks > 0 and suffix and blocks[-1][1] == "1":
        this_suffix = blocks[-1][0]
        blocks = blocks[:-1]
jhoogenboom's avatar
jhoogenboom committed
958
959
960
961

    # Generate prefix/suffix variants.
    length = 0
    variants = []
962
    if prefix != this_prefix:
963
        variants += call_variants(prefix, this_prefix, "prefix")
964
965
        length += len(this_prefix) - len(prefix)
    if suffix != this_suffix:
966
        variants += call_variants(suffix, this_suffix, "suffix")
967
        length += len(this_suffix) - len(suffix)
jhoogenboom's avatar
jhoogenboom committed
968
969
970

    # We are ready to return the allele name of aliases.
    if alias != marker:
971
        return "_".join([library["aliases"][alias]["name"]] + variants)
jhoogenboom's avatar
jhoogenboom committed
972
973
974

    # Compute CE allele number for the other alleles.
    # TODO: perhaps produce a more intelligent name if there is exactly
975
976
    #       1 alias with the same length, or only 1 alias sequence is
    #       contained somewhere within the allele.
jhoogenboom's avatar
jhoogenboom committed
977
    blocknames = []
978
979
    blocksize = library.get("block_length", {}).get(marker, 4)
    length -= library.get("length_adjust", {}).get(marker, 0)
jhoogenboom's avatar
jhoogenboom committed
980
981
982
983
984
985
986
987
988
989
990
991
992
993
    for block in blocks:
        blocknames.append("%s[%s]" % (block[0], block[1]))
        length += len(block[0]) * int(block[1])

    allelename = "CE" + str(length / blocksize)
    if length % blocksize:
        allelename += "." + str(length % blocksize)
    return "_".join([allelename, "".join(blocknames)] + variants)
#convert_sequence_raw_allelename


def ensure_sequence_format(seq, to_format, from_format=None, library=None,
                           marker=None):
    """Convert seq to 'raw', 'tssv', or 'allelename' format."""
jhoogenboom's avatar
jhoogenboom committed
994
    known_formats = ("raw", "tssv", "allelename")
jhoogenboom's avatar
jhoogenboom committed
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
    if to_format not in known_formats:
        raise ValueError("Unknown format '%s', choose from %s" %
                         (to_format, known_formats))
    if from_format is None:
        from_format = detect_sequence_format(seq) if seq else "raw"
    elif from_format not in known_formats:
        raise ValueError("Unknown format '%s', choose from %s" %
                         (from_format, known_formats))

    # No conversion needed?
    if to_format == from_format:
        return seq

    # From TSSV to raw sequence is easy.
    # We'll need a library and marker name for anything else.
    if (library is None or marker is None) and (from_format != "tssv" or
            to_format != "raw"):
        raise ValueError("Sequence needs to be converted from %s to %s, this "
                          "conversion requires a library file" %
                          (from_format, to_format))

    # Perform conversions.
    if from_format == "allelename":
        seq = convert_sequence_allelename_tssv(seq, library, marker)
    if to_format == "tssv":
        if from_format == "raw":
            return convert_sequence_raw_tssv(seq, library, marker)
        return seq
    if from_format != "raw":
        seq = convert_sequence_tssv_raw(seq)
    if to_format == "raw":
        return seq
    return convert_sequence_raw_allelename(seq, library, marker)
#ensure_sequence_format


jhoogenboom's avatar
jhoogenboom committed
1031
1032
def reverse_complement(sequence):
    """Return the reverse complement of the given DNA sequence."""
jhoogenboom's avatar
jhoogenboom committed
1033
    return "".join(COMPL[x] if x in COMPL else x for x in reversed(sequence))
jhoogenboom's avatar
jhoogenboom committed
1034
1035
1036
#reverse_complement


jhoogenboom's avatar
jhoogenboom committed
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
def nnls(A, C, B=None, max_iter=200, min_change=0.0001, debug=False):
    """
    Solve for B in A * B = C in the least squares sense, s.t. B >= 0.

    Hint: call nnls(B.T, C.T).T to solve for A.

    Algorithm has converged if the sum of squared error has decreased
    by less than a factor of min_change in one iteration.  If debug is
    True, print the sum of squared error to sys.stdout after each
    iteration.

    This code was partially adopted from nimfa.methods.factorization.bd.
    """
    import numpy as np
    if B is None:
        B = np.matrix(np.zeros([A.shape[1], C.shape[1]]))
    E = A.T * A
    F = A.T * C
    prev_score = cur_score = sys.float_info.max
    for i in range(max_iter):
        for n in range(B.shape[0]):
            nn = list(range(n)) + list(range(n + 1, B.shape[0]))
            tmp = (F[n, :] - E[n, nn] * B[nn, :]) / E[n, n]
            tmp[np.isnan(tmp)] = 0
            tmp[tmp < 0] = 0
            B[n, :] = tmp
        prev_score = cur_score
        cur_score = np.square(C - A * B).sum()
        score_change = (prev_score-cur_score)/prev_score

        if debug:
            if i:
                print("%4i %15.6f %15.6f %6.2f" % (i, cur_score,
                    prev_score-cur_score, 100*score_change))
            else:
                print("%4i %15.6f" % (i, cur_score))

        if not cur_score or score_change < min_change:
            # We have converged.
            break

    return B
#nnls


jhoogenboom's avatar
jhoogenboom committed
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
def adjust_stats(value, stats=None):
    """
    Adjust the given stats in place with the given observed value and
    return the adjusted stats as well.  If no stats dict is given,
    create a new stats dict with the following initial values:
    {"n": 1, "min": value, "max": value, "mean": value, "m2": 0.0,
     "variance": 0.0}
    """
    value += 0.0
    if not stats:
        return {"n": 1, "min": value, "max": value, "mean": value, "m2": 0.0,
                "variance": 0.0}
    stats["n"] += 1
    delta = value - stats["mean"]
    stats["mean"] += delta / stats["n"]
    stats["m2"] += delta * (value - stats["mean"])
    try:
        stats["variance"] = stats["m2"] / (stats["n"] - 1)
        stats["min"] = min(stats["min"], value)
        stats["max"] = max(stats["max"], value)
    except ZeroDivisionError:
        stats["variance"] = 0
        stats["min"] = value
        stats["max"] = value
    return stats
#adjust_stats


jhoogenboom's avatar
jhoogenboom committed
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
def get_repeat_pattern(seq):
    """Return compiled regular expression that finds repeats of seq."""
    return re.compile("".join(             # For AGAT, one obtains:
        ["(?:" * (len(seq)-1)] +           # (?:(?:(?:
        ["%s)?" % x for x in seq[1:]] +    # G)?A)?T)?
        ["(?:", seq, ")+"] +               # (?AGAT)+
        ["(?:%s" % x for x in seq[:-1]] +  # (?:A(?:G(?:A
        [")?" * (len(seq)-1)]))            # )?)?)?
#get_repeat_pattern


1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
def read_sample_data_file(infile, data, annotation_column=None, seqformat=None,
                          library=None, default_marker=None):
    """Add data from infile to data dict as [marker, allele]=reads."""
    # Get column numbers.
    column_names = infile.readline().rstrip("\r\n").split("\t")
    colid_allele, colid_forward, colid_reverse = \
        get_column_ids(column_names, "allele", "forward", "reverse")

    # Get marker name column if it exists.
    try:
        colid_name = get_column_ids(column_names, "name")
    except:
        colid_name = None

    # Also try to get annotation column if we have one.
    if annotation_column is not None:
        try:
            colid_annotation = get_column_ids(column_names, annotation_column)
        except:
            annotation_column = None

    found_alleles = []
    for line in infile:
        line = line.rstrip("\r\n").split("\t")
        marker = line[colid_name] if colid_name is not None else default_marker
        allele = line[colid_allele] if seqformat is None \
jhoogenboom's avatar
jhoogenboom committed
1147
1148
            else ensure_sequence_format(line[colid_allele], seqformat,
                                        library=library, marker=marker)
1149
1150
        if (annotation_column is not None and
                line[colid_annotation].startswith("ALLELE")):
jhoogenboom's avatar
jhoogenboom committed
1151
            found_alleles.append((marker, allele))
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
        data[marker, allele] = map(int,
            (line[colid_forward], line[colid_reverse]))

    return found_alleles
#read_sample_data_file


def reduce_read_counts(data, limit_reads):
    sum_reads = 0
    for markerallele in data:
        sum_reads += sum(data[markerallele])
    if sum_reads <= limit_reads:
        return

    remove = sorted(random.sample(xrange(sum_reads), sum_reads - limit_reads))
    i = 0
    seen = 0
    while i < len(remove) and seen > remove[i]:
        # Skip the reads filtered out above.
        i += 1
    for markerallele in data:
        for direction in (0, 1):
            seen += data[markerallele][direction]
            while i < len(remove) and seen > remove[i]:
                data[markerallele][direction] -= 1
                i += 1
#reduce_read_counts


def get_sample_data(tags_to_files, callback, allelelist=None,
                    annotation_column=None, seqformat=None, library=None,
                    marker=None, homozygotes=False, limit_reads=sys.maxint,
                    drop_samples=0):
    if drop_samples:
        sample_tags = tags_to_files.keys()
        for tag in random.sample(xrange(len(sample_tags)),
                                 int(len(sample_tags) * drop_samples)):
            del tags_to_files[sample_tags[tag]]

    for tag in tags_to_files:
        data = {}
        alleles = set()
        for infile in tags_to_files[tag]:
1195
            infile = sys.stdin if infile == "-" else open(infile, "r")
1196
1197
            alleles.update(read_sample_data_file(infile, data,
                annotation_column, seqformat, library, marker))
1198
1199
            if infile != sys.stdin:
                infile.close()
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
        if limit_reads < sys.maxint:
            reduce_read_counts(data, limit_reads)
        if allelelist is not None:
            if tag not in allelelist:
                allelelist[tag] = {}
            for markerx, allele in alleles:
                if markerx not in allelelist[tag]:
                    allelelist[tag][markerx] = set()
                allelelist[tag][markerx].add(allele)
            if marker:
                if marker in allelelist[tag]:
                    allelelist[tag] = {marker: allelelist[tag][marker]}
                else:
                    allelelist[tag] = {}
            if homozygotes:
jhoogenboom's avatar
jhoogenboom committed
1215
1216
1217
                for markerx in allelelist[tag].keys():
                    if len(allelelist[tag][markerx]) > 1:
                        del allelelist[tag][markerx]
1218
1219
1220
1221
        callback(tag, data)
#get_sample_data


jhoogenboom's avatar
jhoogenboom committed
1222
1223
1224
1225
1226
1227
1228
def get_column_ids(column_names, *names):
    """Find all names in column_names and return their indices."""
    result = []
    for name in names:
        try:
            result.append(column_names.index(name))
        except ValueError:
jhoogenboom's avatar
jhoogenboom committed
1229
            raise ValueError("Column not found in input file: %s" % name)
jhoogenboom's avatar
jhoogenboom committed
1230
1231
1232
1233
1234
1235
    if len(result) == 1:
        return result[0]
    return tuple(result)
#get_column_ids


jhoogenboom's avatar
jhoogenboom committed
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
def parse_allelelist(allelelist, convert=None, library=None):
    """Read allele list from open file handle."""
    column_names = allelelist.readline().rstrip("\r\n").split("\t")
    colid_sample, colid_marker, colid_allele = get_column_ids(column_names,
        "sample", "marker", "allele")
    alleles = {}
    for line in allelelist:
        line = line.rstrip("\r\n").split("\t")
        sample = line[colid_sample]
        marker = line[colid_marker]
        allele = line[colid_allele]
        if convert is not None:
            allele = ensure_sequence_format(allele, convert, library=library,
                                            marker=marker)
        if sample not in alleles:
            alleles[sample] = {}
        if marker not in alleles[sample]:
            alleles[sample][marker] = set()
        alleles[sample][marker].add(allele)
    return alleles
#parse_allelelist


jhoogenboom's avatar
jhoogenboom committed
1259
1260
1261
1262
1263
1264
1265
1266
1267
def pos_int_arg(value):
    """Convert str to int, raise ArgumentTypeError if not positive."""
    if not value.isdigit() or not int(value):
        raise argparse.ArgumentTypeError(
            "invalid positive int value: '%s'" % value)
    return int(value)
#pos_int_arg


1268
1269
1270
1271
1272
1273
1274
1275
1276
def regex_arg(value):
    """Compile value into a regular expression."""
    try:
        return re.compile(value)
    except re.error as err:
        raise argparse.ArgumentTypeError(err)
#regex_arg


jhoogenboom's avatar
jhoogenboom committed
1277
def add_allele_detection_args(parser):
jhoogenboom's avatar
jhoogenboom committed
1278
1279
    group = parser.add_argument_group("allele detection options")
    group.add_argument('-a', '--allelelist', metavar="ALLELEFILE",
jhoogenboom's avatar
jhoogenboom committed
1280
        type=argparse.FileType('r'),
jhoogenboom's avatar
jhoogenboom committed
1281
1282
1283
        help="file containing a list of the true alleles of each sample "
             "(e.g., obtained from allelefinder)")
    group.add_argument('-c', '--annotation-column', metavar="COLNAME",
jhoogenboom's avatar
jhoogenboom committed
1284
1285
1286
1287
1288
        help="name of a column in the sample files, which contains a value "
             "beginning with 'ALLELE' for the true alleles of the sample")
#add_allele_detection_args


1289
1290
def add_random_subsampling_args(parser):
    group = parser.add_argument_group("random subsampling options (advanced)")
jhoogenboom's avatar
jhoogenboom committed
1291
    group.add_argument('-Q', '--limit-reads', metavar="N", type=pos_int_arg,
1292
1293
1294
1295
1296
1297
        default=sys.maxint,
        help="simulate lower sequencing depth by randomly dropping reads down "
             "to this maximum total number of reads for each sample")
    group.add_argument('-x', '--drop-samples', metavar="N", type=float,
        default=0, help="randomly drop this fraction of input samples")
#add_random_subsampling_args
jhoogenboom's avatar
jhoogenboom committed
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312


def add_sequence_format_args(parser, default_format=None, force=False):
    group = parser.add_argument_group("sequence format options")
    if force:
        group.set_defaults(sequence_format=default_format)
    else:
        group.add_argument('-F', '--sequence-format', metavar="FORMAT",
            choices=("raw", "tssv", "allelename"),
            default=default_format,
            help="convert sequences to the specified format: one of "
                 "%(choices)s (default: " + (
                 "no conversion" if default_format is None else default_format)
                 + ")")
    group.add_argument('-l', '--library', metavar="LIBRARY",
1313
        type=parse_library,
jhoogenboom's avatar
jhoogenboom committed
1314
1315
1316
1317
        help="library file for sequence format conversion")
#add_sequence_format_args


1318
1319
1320
1321
1322
1323
def add_input_output_args(parser, single_in=False, batch_support=False,
                          report_out=False):
    """Add arguments for opening sample files to the given parser."""
    # Input file options group.
    if not single_in:
        parser.add_argument('infiles', nargs='*', metavar="FILE",
1324
            default=["-"],
1325
1326
1327
1328
            help="the sample data file(s) to process (default: read from "
                 "stdin)")
    elif not batch_support:
        parser.add_argument('infile', nargs='?', metavar="IN",
1329
            default="-",
1330
1331
1332
1333
1334
            help="the sample data file to process (default: read from stdin)")
    else:
        mutex = parser.add_argument_group(
                    "input file options").add_mutually_exclusive_group()
        mutex.add_argument('infile', nargs='?', metavar="IN",
1335
            default="-",
1336
1337
1338
            help="single sample data file to process (default: read from "
                 "stdin)")
        mutex.add_argument("-i", "--input", dest="infiles", nargs="+",
1339
            metavar="IN",
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
            help="multiple sample data files to process (use with "
                 "-o/--output)")

    # Output file options group.
    group = parser.add_argument_group("output file options")
    if batch_support and single_in:
        mutex = group.add_mutually_exclusive_group()
        mutex.add_argument('outfile', nargs='?', metavar="OUT",
            default=sys.stdout,
            help="the file to write the output to (default: write to stdout)")
        mutex.add_argument('-o', '--output', dest="outfiles", nargs="+",
            metavar="OUT",
            help="list of names of output files to match with input files "
                 "specified with -i/--input, or a format string to construct "
                 "file names from sample tags; e.g., the default value is "
                 "'\\1-%s.out', which expands to 'sampletag-%s.out'" %
                    ((parser.prog.rsplit(" ", 1)[-1],)*2))
    elif batch_support:
        group.add_argument('-o', '--output', dest="outfiles", nargs="+",
            metavar="OUT",
            default=[sys.stdout],
            help="a single file name to write all output to (default: write "
                 "to stdout) OR a list of names of output files to match with "
                 "input files OR a format string to construct file names from "
                 "sample tags; e.g., the value '\\1-%s.out' expands to "
                 "'sampletag-%s.out'" % ((parser.prog.rsplit(" ", 1)[-1],)*2))
    else:
        group.add_argument('-o', '--output', dest="outfile", metavar="FILE",
            type=argparse.FileType('w'),
            default=sys.stdout,
            help="file to write output to (default: write to stdout)")
    if report_out:
jhoogenboom's avatar
jhoogenboom committed
1372
        group.add_argument('-R', '--report', metavar="FILE",
1373
1374
1375
1376
1377
1378
            type=argparse.FileType('w'),
            default=sys.stderr,
            help="file to write a report to (default: write to stderr)")

    # Sample tag parsing options group.
    group = parser.add_argument_group("sample tag parsing options")
1379
    group.add_argument('-e', '--tag-expr', metavar="REGEX", type=regex_arg,
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
        default=DEF_TAG_EXPR,
        help="regular expression that captures (using one or more capturing "
             "groups) the sample tags from the file names; by default, the "
             "entire file name except for its extension (if any) is captured")