scripts/last-dotplot
author Martin C. Frith
Thu Jun 15 13:42:42 2017 +0900 (2017-06-15)
changeset 866 5182d8528ce9
parent 863 6a4915d5b5cb
child 875 592295375eb1
permissions -rwxr-xr-x
Change last-dotplot's verbosity
Martin@1
     1
#! /usr/bin/env python
Martin@1
     2
Martin@272
     3
# Read pair-wise alignments in MAF or LAST tabular format: write an
Martin@272
     4
# "Oxford grid", a.k.a. dotplot.
Martin@1
     5
Martin@1
     6
# TODO: Currently, pixels with zero aligned nt-pairs are white, and
Martin@1
     7
# pixels with one or more aligned nt-pairs are black.  This can look
Martin@1
     8
# too crowded for large genome alignments.  I tried shading each pixel
Martin@1
     9
# according to the number of aligned nt-pairs within it, but the
Martin@1
    10
# result is too faint.  How can this be done better?
Martin@1
    11
Martin@844
    12
import fnmatch, itertools, optparse, os, re, sys
Martin@475
    13
Martin@475
    14
# Try to make PIL/PILLOW work:
Martin@475
    15
try: from PIL import Image, ImageDraw, ImageFont, ImageColor
Martin@475
    16
except ImportError: import Image, ImageDraw, ImageFont, ImageColor
Martin@1
    17
Martin@844
    18
def myOpen(fileName):  # faster than fileinput
Martin@844
    19
    if fileName == "-":
Martin@844
    20
        return sys.stdin
Martin@844
    21
    return open(fileName)
Martin@844
    22
Martin@644
    23
def warn(message):
Martin@866
    24
    if opts.verbose:
Martin@866
    25
        prog = os.path.basename(sys.argv[0])
Martin@866
    26
        sys.stderr.write(prog + ": " + message + "\n")
Martin@644
    27
Martin@840
    28
def croppedBlocks(blocks, range1, range2):
Martin@840
    29
    cropBeg1, cropEnd1 = range1
Martin@840
    30
    cropBeg2, cropEnd2 = range2
Martin@840
    31
    if blocks[0][0] < 0: cropBeg1, cropEnd1 = -cropEnd1, -cropBeg1
Martin@840
    32
    if blocks[0][1] < 0: cropBeg2, cropEnd2 = -cropEnd2, -cropBeg2
Martin@840
    33
    for beg1, beg2, size in blocks:
Martin@840
    34
        b1 = max(cropBeg1, beg1)
Martin@840
    35
        e1 = min(cropEnd1, beg1 + size)
Martin@840
    36
        if b1 >= e1: continue
Martin@840
    37
        offset = beg2 - beg1
Martin@840
    38
        b2 = max(cropBeg2, b1 + offset)
Martin@840
    39
        e2 = min(cropEnd2, e1 + offset)
Martin@840
    40
        if b2 >= e2: continue
Martin@840
    41
        yield b2 - offset, b2, e2 - b2
Martin@840
    42
Martin@482
    43
def tabBlocks(beg1, beg2, blocks):
Martin@482
    44
    '''Get the gapless blocks of an alignment, from LAST tabular format.'''
Martin@482
    45
    for i in blocks.split(","):
Martin@482
    46
        if ":" in i:
Martin@482
    47
            x, y = i.split(":")
Martin@482
    48
            beg1 += int(x)
Martin@482
    49
            beg2 += int(y)
Martin@482
    50
        else:
Martin@482
    51
            size = int(i)
Martin@482
    52
            yield beg1, beg2, size
Martin@482
    53
            beg1 += size
Martin@482
    54
            beg2 += size
Martin@272
    55
Martin@482
    56
def mafBlocks(beg1, beg2, seq1, seq2):
Martin@482
    57
    '''Get the gapless blocks of an alignment, from MAF format.'''
Martin@482
    58
    size = 0
Martin@482
    59
    for x, y in itertools.izip(seq1, seq2):
Martin@482
    60
        if x == "-":
Martin@482
    61
            if size:
Martin@482
    62
                yield beg1, beg2, size
Martin@482
    63
                beg1 += size
Martin@482
    64
                beg2 += size
Martin@482
    65
                size = 0
Martin@482
    66
            beg2 += 1
Martin@482
    67
        elif y == "-":
Martin@482
    68
            if size:
Martin@482
    69
                yield beg1, beg2, size
Martin@482
    70
                beg1 += size
Martin@482
    71
                beg2 += size
Martin@482
    72
                size = 0
Martin@482
    73
            beg1 += 1
Martin@272
    74
        else:
Martin@482
    75
            size += 1
Martin@482
    76
    if size: yield beg1, beg2, size
Martin@272
    77
Martin@482
    78
def alignmentInput(lines):
Martin@482
    79
    '''Get alignments and sequence lengths, from MAF or tabular format.'''
Martin@482
    80
    mafCount = 0
Martin@272
    81
    for line in lines:
Martin@272
    82
        w = line.split()
Martin@272
    83
        if line[0].isdigit():  # tabular format
Martin@482
    84
            chr1, beg1, seqlen1 = w[1], int(w[2]), int(w[5])
Martin@482
    85
            if w[4] == "-": beg1 -= seqlen1
Martin@482
    86
            chr2, beg2, seqlen2 = w[6], int(w[7]), int(w[10])
Martin@482
    87
            if w[9] == "-": beg2 -= seqlen2
Martin@847
    88
            blocks = tabBlocks(beg1, beg2, w[11])
Martin@482
    89
            yield chr1, seqlen1, chr2, seqlen2, blocks
Martin@272
    90
        elif line[0] == "s":  # MAF format
Martin@482
    91
            if mafCount == 0:
Martin@482
    92
                chr1, beg1, seqlen1, seq1 = w[1], int(w[2]), int(w[5]), w[6]
Martin@482
    93
                if w[4] == "-": beg1 -= seqlen1
Martin@482
    94
                mafCount = 1
Martin@482
    95
            else:
Martin@482
    96
                chr2, beg2, seqlen2, seq2 = w[1], int(w[2]), int(w[5]), w[6]
Martin@482
    97
                if w[4] == "-": beg2 -= seqlen2
Martin@847
    98
                blocks = mafBlocks(beg1, beg2, seq1, seq2)
Martin@482
    99
                yield chr1, seqlen1, chr2, seqlen2, blocks
Martin@482
   100
                mafCount = 0
Martin@272
   101
Martin@840
   102
def seqRangeFromText(text):
Martin@840
   103
    if ":" in text:
Martin@840
   104
        pattern, interval = text.rsplit(":", 1)
Martin@840
   105
        if "-" in interval:
Martin@840
   106
            beg, end = interval.rsplit("-", 1)
Martin@840
   107
            return pattern, int(beg), int(end)  # beg may be negative
Martin@840
   108
    return text, 0, sys.maxsize
Martin@840
   109
Martin@840
   110
def rangeFromSeqName(seqRanges, name, seqLen):
Martin@840
   111
    if not seqRanges: return 0, seqLen
Martin@651
   112
    base = name.split(".")[-1]  # allow for names like hg19.chr7
Martin@840
   113
    for pat, beg, end in seqRanges:
Martin@840
   114
        if fnmatch.fnmatchcase(name, pat) or fnmatch.fnmatchcase(base, pat):
Martin@840
   115
            return max(beg, 0), min(end, seqLen)
Martin@844
   116
    return None
Martin@651
   117
Martin@851
   118
def updateSeqs(isTrim, seqNames, seqLimits, seqName, seqRange, blocks, index):
Martin@851
   119
    if seqName not in seqLimits:
Martin@851
   120
        seqNames.append(seqName)
Martin@839
   121
    if isTrim:
Martin@839
   122
        beg = blocks[0][index]
Martin@839
   123
        end = blocks[-1][index] + blocks[-1][2]
Martin@839
   124
        if beg < 0: beg, end = -end, -beg
Martin@839
   125
        if seqName in seqLimits:
Martin@839
   126
            b, e = seqLimits[seqName]
Martin@839
   127
            seqLimits[seqName] = min(b, beg), max(e, end)
Martin@839
   128
        else:
Martin@839
   129
            seqLimits[seqName] = beg, end
Martin@839
   130
    else:
Martin@840
   131
        seqLimits[seqName] = seqRange
Martin@839
   132
Martin@651
   133
def readAlignments(fileName, opts):
Martin@839
   134
    '''Get alignments and sequence limits, from MAF or tabular format.'''
Martin@840
   135
    seqRanges1 = map(seqRangeFromText, opts.seq1)
Martin@840
   136
    seqRanges2 = map(seqRangeFromText, opts.seq2)
Martin@840
   137
Martin@482
   138
    alignments = []
Martin@851
   139
    seqNames1 = []
Martin@851
   140
    seqNames2 = []
Martin@839
   141
    seqLimits1 = {}
Martin@839
   142
    seqLimits2 = {}
Martin@844
   143
    lines = myOpen(fileName)
Martin@838
   144
    for seqName1, seqLen1, seqName2, seqLen2, blocks in alignmentInput(lines):
Martin@840
   145
        range1 = rangeFromSeqName(seqRanges1, seqName1, seqLen1)
Martin@844
   146
        if not range1: continue
Martin@840
   147
        range2 = rangeFromSeqName(seqRanges2, seqName2, seqLen2)
Martin@844
   148
        if not range2: continue
Martin@847
   149
        b = list(croppedBlocks(list(blocks), range1, range2))
Martin@840
   150
        if not b: continue
Martin@840
   151
        aln = seqName1, seqName2, b
Martin@482
   152
        alignments.append(aln)
Martin@851
   153
        updateSeqs(opts.trim1, seqNames1, seqLimits1, seqName1, range1, b, 0)
Martin@851
   154
        updateSeqs(opts.trim2, seqNames2, seqLimits2, seqName2, range2, b, 1)
Martin@851
   155
    return alignments, seqNames1, seqNames2, seqLimits1, seqLimits2
Martin@1
   156
Martin@1
   157
def natural_sort_key(my_string):
Martin@1
   158
    '''Return a sort key for "natural" ordering, e.g. chr9 < chr10.'''
Martin@1
   159
    parts = re.split(r'(\d+)', my_string)
Martin@1
   160
    parts[1::2] = map(int, parts[1::2])
Martin@1
   161
    return parts
Martin@1
   162
Martin@647
   163
def get_text_sizes(my_strings, font, fontsize, image_mode):
Martin@1
   164
    '''Get widths & heights, in pixels, of some strings.'''
Martin@647
   165
    if fontsize == 0: return [(0, 0) for i in my_strings]
Martin@1
   166
    image_size = 1, 1
Martin@134
   167
    im = Image.new(image_mode, image_size)
Martin@1
   168
    draw = ImageDraw.Draw(im)
Martin@1
   169
    return [draw.textsize(i, font=font) for i in my_strings]
Martin@1
   170
Martin@846
   171
def sizeText(size):
Martin@846
   172
    suffixes = "bp", "kb", "Mb", "Gb"
Martin@846
   173
    for i, x in enumerate(suffixes):
Martin@846
   174
        j = 10 ** (i * 3)
Martin@846
   175
        if size < j * 10:
Martin@846
   176
            return "%.2g" % (1.0 * size / j) + x
Martin@846
   177
        if size < j * 1000 or i == len(suffixes) - 1:
Martin@846
   178
            return "%.0f" % (1.0 * size / j) + x
Martin@846
   179
Martin@846
   180
def seqNameAndSizeText(seqName, seqSize):
Martin@846
   181
    return seqName + ": " + sizeText(seqSize)
Martin@846
   182
Martin@851
   183
def getSeqInfo(sortOpt, seqNames, seqLimits,
Martin@851
   184
               font, fontsize, image_mode, isShowSize):
Martin@1
   185
    '''Return miscellaneous information about the sequences.'''
Martin@851
   186
    if sortOpt == 1:
Martin@851
   187
        seqNames.sort(key=natural_sort_key)
Martin@850
   188
    seqSizes = [seqLimits[i][1] - seqLimits[i][0] for i in seqNames]
Martin@866
   189
    for i in seqNames:
Martin@866
   190
        r = seqLimits[i]
Martin@866
   191
        out = i, str(r[0]), str(r[1])
Martin@866
   192
        warn("\t".join(out))
Martin@866
   193
    warn("")
Martin@851
   194
    if sortOpt == 2:
Martin@851
   195
        seqRecords = sorted(zip(seqSizes, seqNames), reverse=True)
Martin@851
   196
        seqSizes = [i[0] for i in seqRecords]
Martin@851
   197
        seqNames = [i[1] for i in seqRecords]
Martin@850
   198
    if isShowSize:
Martin@850
   199
        seqLabels = map(seqNameAndSizeText, seqNames, seqSizes)
Martin@850
   200
    else:
Martin@850
   201
        seqLabels = seqNames
Martin@846
   202
    labelSizes = get_text_sizes(seqLabels, font, fontsize, image_mode)
Martin@846
   203
    margin = max(zip(*labelSizes)[1])  # maximum text height
Martin@850
   204
    return seqNames, seqSizes, seqLabels, labelSizes, margin
Martin@1
   205
Martin@1
   206
def div_ceil(x, y):
Martin@1
   207
    '''Return x / y rounded up.'''
Martin@1
   208
    q, r = divmod(x, y)
Martin@1
   209
    return q + (r != 0)
Martin@1
   210
Martin@645
   211
def get_bp_per_pix(seq_sizes, pix_tween_seqs, pix_limit):
Martin@1
   212
    '''Get the minimum bp-per-pixel that fits in the size limit.'''
Martin@1
   213
    seq_num = len(seq_sizes)
Martin@1
   214
    seq_pix_limit = pix_limit - pix_tween_seqs * (seq_num - 1)
Martin@1
   215
    if seq_pix_limit < seq_num:
Martin@649
   216
        raise Exception("can't fit the image: too many sequences?")
Martin@863
   217
    negLimit = -seq_pix_limit
Martin@863
   218
    negBpPerPix = sum(seq_sizes) // negLimit
Martin@863
   219
    while True:
Martin@863
   220
        if sum(i // negBpPerPix for i in seq_sizes) >= negLimit:
Martin@863
   221
            return -negBpPerPix
Martin@863
   222
        negBpPerPix -= 1
Martin@1
   223
Martin@1
   224
def get_seq_starts(seq_pix, pix_tween_seqs, margin):
Martin@1
   225
    '''Get the start pixel for each sequence.'''
Martin@1
   226
    seq_starts = []
Martin@1
   227
    pix_tot = margin - pix_tween_seqs
Martin@1
   228
    for i in seq_pix:
Martin@1
   229
        pix_tot += pix_tween_seqs
Martin@1
   230
        seq_starts.append(pix_tot)
Martin@1
   231
        pix_tot += i
Martin@1
   232
    return seq_starts
Martin@1
   233
Martin@645
   234
def get_pix_info(seq_sizes, bp_per_pix, pix_tween_seqs, margin):
Martin@1
   235
    '''Return pixel information about the sequences.'''
Martin@1
   236
    seq_pix = [div_ceil(i, bp_per_pix) for i in seq_sizes]
Martin@1
   237
    seq_starts = get_seq_starts(seq_pix, pix_tween_seqs, margin)
Martin@1
   238
    tot_pix = seq_starts[-1] + seq_pix[-1]
Martin@1
   239
    return seq_pix, seq_starts, tot_pix
Martin@1
   240
Martin@835
   241
def drawLineForward(hits, width, bp_per_pix, beg1, beg2, size):
Martin@639
   242
    while True:
Martin@639
   243
        q1, r1 = divmod(beg1, bp_per_pix)
Martin@639
   244
        q2, r2 = divmod(beg2, bp_per_pix)
Martin@835
   245
        hits[q2 * width + q1] |= 1
Martin@639
   246
        next_pix = min(bp_per_pix - r1, bp_per_pix - r2)
Martin@639
   247
        if next_pix >= size: break
Martin@639
   248
        beg1 += next_pix
Martin@639
   249
        beg2 += next_pix
Martin@639
   250
        size -= next_pix
Martin@639
   251
Martin@835
   252
def drawLineReverse(hits, width, bp_per_pix, beg1, beg2, size):
Martin@639
   253
    beg2 = -1 - beg2
Martin@639
   254
    while True:
Martin@639
   255
        q1, r1 = divmod(beg1, bp_per_pix)
Martin@639
   256
        q2, r2 = divmod(beg2, bp_per_pix)
Martin@835
   257
        hits[q2 * width + q1] |= 2
Martin@639
   258
        next_pix = min(bp_per_pix - r1, r2 + 1)
Martin@639
   259
        if next_pix >= size: break
Martin@639
   260
        beg1 += next_pix
Martin@639
   261
        beg2 -= next_pix
Martin@639
   262
        size -= next_pix
Martin@639
   263
Martin@836
   264
def alignmentPixels(width, height, alignments, bp_per_pix, origins1, origins2):
Martin@640
   265
    hits = [0] * (width * height)  # the image data
Martin@640
   266
    for seq1, seq2, blocks in alignments:
Martin@836
   267
        ori1 = origins1[seq1]
Martin@836
   268
        ori2 = origins2[seq2]
Martin@640
   269
        for beg1, beg2, size in blocks:
Martin@640
   270
            if beg1 < 0:
Martin@640
   271
                beg1 = -(beg1 + size)
Martin@640
   272
                beg2 = -(beg2 + size)
Martin@640
   273
            if beg2 >= 0:
Martin@835
   274
                drawLineForward(hits, width, bp_per_pix,
Martin@835
   275
                                beg1 + ori1, beg2 + ori2, size)
Martin@640
   276
            else:
Martin@835
   277
                drawLineReverse(hits, width, bp_per_pix,
Martin@835
   278
                                beg1 + ori1, beg2 - ori2, size)
Martin@640
   279
    return hits
Martin@1
   280
Martin@650
   281
def expandedSeqDict(seqDict):
Martin@650
   282
    '''Allow lookup by short sequence names, e.g. chr7 as well as hg19.chr7.'''
Martin@861
   283
    newDict = seqDict.copy()
Martin@650
   284
    for name, x in seqDict.items():
Martin@861
   285
        if "." in name:
Martin@861
   286
            base = name.split(".")[-1]
Martin@861
   287
            if base in newDict:  # an ambiguous case was found:
Martin@861
   288
                return seqDict   # so give up completely
Martin@861
   289
            newDict[base] = x
Martin@650
   290
    return newDict
Martin@650
   291
Martin@845
   292
def readBed(fileName, seqLimits):
Martin@845
   293
    if not fileName: return
Martin@845
   294
    for line in myOpen(fileName):
Martin@845
   295
        w = line.split()
Martin@847
   296
        if not w: continue
Martin@845
   297
        seqName = w[0]
Martin@845
   298
        if seqName not in seqLimits: continue
Martin@845
   299
        beg = int(w[1])
Martin@845
   300
        end = int(w[2])
Martin@857
   301
        layer = 900
Martin@856
   302
        color = "#ffe4ff"
Martin@858
   303
        if len(w) > 4:
Martin@858
   304
            if w[4] != ".":
Martin@858
   305
                layer = float(w[4])
Martin@858
   306
            if len(w) > 5:
Martin@858
   307
                if len(w) > 8 and w[8].count(",") == 2:
Martin@858
   308
                    color = "rgb(" + w[8] + ")"
Martin@858
   309
                elif w[5] == "+":
Martin@858
   310
                    color = "#fff4f4"
Martin@858
   311
                elif w[5] == "-":
Martin@858
   312
                    color = "#f4f4ff"
Martin@859
   313
        yield layer, color, seqName, beg, end
Martin@845
   314
Martin@860
   315
def commaSeparatedInts(text):
Martin@860
   316
    return map(int, text.rstrip(",").split(","))
Martin@860
   317
Martin@860
   318
def readGenePred(opts, fileName, seqLimits):
Martin@860
   319
    if not fileName: return
Martin@860
   320
    for line in myOpen(fileName):
Martin@860
   321
        fields = line.split()
Martin@860
   322
        if not fields: continue
Martin@860
   323
        if fields[2] not in "+-": fields = fields[1:]
Martin@860
   324
        seqName = fields[1]
Martin@860
   325
        if seqName not in seqLimits: continue
Martin@860
   326
        #strand = fields[2]
Martin@860
   327
        cdsBeg = int(fields[5])
Martin@860
   328
        cdsEnd = int(fields[6])
Martin@860
   329
        exonBegs = commaSeparatedInts(fields[8])
Martin@860
   330
        exonEnds = commaSeparatedInts(fields[9])
Martin@860
   331
        for beg, end in zip(exonBegs, exonEnds):
Martin@860
   332
            yield 300, opts.exon_color, seqName, beg, end
Martin@860
   333
            b = max(beg, cdsBeg)
Martin@860
   334
            e = min(end, cdsEnd)
Martin@860
   335
            if b < e: yield 400, opts.cds_color, seqName, b, e
Martin@860
   336
Martin@860
   337
def readRmsk(fileName, seqLimits):
Martin@860
   338
    if not fileName: return
Martin@860
   339
    for line in myOpen(fileName):
Martin@860
   340
        fields = line.split()
Martin@860
   341
        if len(fields) == 17:  # rmsk.txt
Martin@860
   342
            seqName = fields[5]
Martin@860
   343
            if seqName not in seqLimits: continue  # do this ASAP for speed
Martin@860
   344
            beg = int(fields[6])
Martin@860
   345
            end = int(fields[7])
Martin@860
   346
            strand = fields[9]
Martin@860
   347
            repeatClass = fields[11]
Martin@860
   348
        elif len(fields) == 15:  # .out
Martin@860
   349
            seqName = fields[4]
Martin@860
   350
            if seqName not in seqLimits: continue
Martin@860
   351
            beg = int(fields[5]) - 1
Martin@860
   352
            end = int(fields[6])
Martin@860
   353
            strand = fields[8]
Martin@860
   354
            repeatClass = fields[10]
Martin@860
   355
        else:
Martin@860
   356
            continue
Martin@860
   357
        if repeatClass in ("Low_complexity", "Simple_repeat"):
Martin@860
   358
            yield 200, "#ffe4ff", seqName, beg, end
Martin@860
   359
        elif strand == "+":
Martin@860
   360
            yield 100, "#fff4f4", seqName, beg, end
Martin@860
   361
        else:
Martin@860
   362
            yield 100, "#f4f4ff", seqName, beg, end
Martin@860
   363
Martin@650
   364
def isExtraFirstGapField(fields):
Martin@650
   365
    return fields[4].isdigit()
Martin@650
   366
Martin@857
   367
def readGaps(opts, fileName, seqLimits):
Martin@650
   368
    '''Read locations of unsequenced gaps, from an agp or gap file.'''
Martin@650
   369
    if not fileName: return
Martin@844
   370
    for line in myOpen(fileName):
Martin@650
   371
        w = line.split()
Martin@650
   372
        if not w or w[0][0] == "#": continue
Martin@650
   373
        if isExtraFirstGapField(w): w = w[1:]
Martin@650
   374
        if w[4] not in "NU": continue
Martin@650
   375
        seqName = w[0]
Martin@839
   376
        if seqName not in seqLimits: continue
Martin@650
   377
        end = int(w[2])
Martin@650
   378
        beg = end - int(w[5])  # zero-based coordinate
Martin@857
   379
        if w[7] == "yes":
Martin@859
   380
            yield 3000, opts.bridged_color, seqName, beg, end
Martin@857
   381
        else:
Martin@859
   382
            yield 2000, opts.unbridged_color, seqName, beg, end
Martin@650
   383
Martin@859
   384
def bedBoxes(beds, seqLimits, origins, margin, edge, isTop, bpPerPix):
Martin@857
   385
    for layer, color, seqName, beg, end in beds:
Martin@859
   386
        cropBeg, cropEnd = seqLimits[seqName]
Martin@859
   387
        beg = max(beg, cropBeg)
Martin@859
   388
        end = min(end, cropEnd)
Martin@859
   389
        if beg >= end: continue
Martin@845
   390
        ori = origins[seqName]
Martin@857
   391
        if layer <= 1000:
Martin@857
   392
            # include partly-covered pixels
Martin@857
   393
            b = (ori + beg) // bpPerPix
Martin@857
   394
            e = div_ceil(ori + end, bpPerPix)
Martin@857
   395
        else:
Martin@857
   396
            # exclude partly-covered pixels
Martin@857
   397
            b = div_ceil(ori + beg, bpPerPix)
Martin@857
   398
            e = (ori + end) // bpPerPix
Martin@857
   399
            if e <= b: continue
Martin@857
   400
        if isTop:
Martin@857
   401
            box = b, margin, e, edge
Martin@857
   402
        else:
Martin@857
   403
            box = margin, b, edge, e
Martin@857
   404
        yield layer, color, box
Martin@845
   405
Martin@857
   406
def drawAnnotations(im, boxes):
Martin@857
   407
    # xxx use partial transparency for different-color overlaps?
Martin@857
   408
    for layer, color, box in boxes:
Martin@650
   409
        im.paste(color, box)
Martin@650
   410
Martin@1
   411
def make_label(text, text_size, range_start, range_size):
Martin@1
   412
    '''Return an axis label with endpoint & sort-order information.'''
Martin@1
   413
    text_width  = text_size[0]
Martin@1
   414
    label_start = range_start + (range_size - text_width) // 2
Martin@1
   415
    label_end   = label_start + text_width
Martin@1
   416
    sort_key    = text_width - range_size
Martin@1
   417
    return sort_key, label_start, label_end, text
Martin@1
   418
Martin@645
   419
def get_nonoverlapping_labels(labels, label_space):
Martin@1
   420
    '''Get a subset of non-overlapping axis labels, greedily.'''
Martin@1
   421
    nonoverlapping_labels = []
Martin@1
   422
    for i in labels:
Martin@28
   423
        if True not in [i[1] < j[2] + label_space and j[1] < i[2] + label_space
Martin@28
   424
                        for j in nonoverlapping_labels]:
Martin@1
   425
            nonoverlapping_labels.append(i)
Martin@1
   426
    return nonoverlapping_labels
Martin@1
   427
Martin@837
   428
def get_axis_image(seqNames, name_sizes, seq_starts, seq_pix,
Martin@647
   429
                   font, image_mode, opts):
Martin@1
   430
    '''Make an image of axis labels.'''
Martin@1
   431
    min_pos = seq_starts[0]
Martin@1
   432
    max_pos = seq_starts[-1] + seq_pix[-1]
Martin@28
   433
    height = max(zip(*name_sizes)[1])
Martin@847
   434
    labels = map(make_label, seqNames, name_sizes, seq_starts, seq_pix)
Martin@1
   435
    labels = [i for i in labels if i[1] >= min_pos and i[2] <= max_pos]
Martin@1
   436
    labels.sort()
Martin@646
   437
    labels = get_nonoverlapping_labels(labels, opts.label_space)
Martin@1
   438
    image_size = max_pos, height
Martin@852
   439
    im = Image.new(image_mode, image_size, opts.border_color)
Martin@1
   440
    draw = ImageDraw.Draw(im)
Martin@1
   441
    for i in labels:
Martin@1
   442
        position = i[1], 0
Martin@646
   443
        draw.text(position, i[3], font=font, fill=opts.text_color)
Martin@1
   444
    return im
Martin@1
   445
Martin@839
   446
def seqOrigins(seqNames, seq_starts, seqLimits, bp_per_pix):
Martin@837
   447
    for i, j in zip(seqNames, seq_starts):
Martin@839
   448
        yield i, bp_per_pix * j - seqLimits[i][0]
Martin@836
   449
Martin@648
   450
def lastDotplot(opts, args):
Martin@643
   451
    if opts.fontfile:  font = ImageFont.truetype(opts.fontfile, opts.fontsize)
Martin@643
   452
    else:              font = ImageFont.load_default()
Martin@641
   453
Martin@643
   454
    image_mode = 'RGB'
Martin@643
   455
    forward_color = ImageColor.getcolor(opts.forwardcolor, image_mode)
Martin@643
   456
    reverse_color = ImageColor.getcolor(opts.reversecolor, image_mode)
Martin@643
   457
    zipped_colors = zip(forward_color, reverse_color)
Martin@643
   458
    overlap_color = tuple([(i + j) // 2 for i, j in zipped_colors])
Martin@641
   459
Martin@644
   460
    warn("reading alignments...")
Martin@851
   461
    alignmentInfo = readAlignments(args[0], opts)
Martin@851
   462
    alignments, seqNames1, seqNames2, seqLimits1, seqLimits2 = alignmentInfo
Martin@644
   463
    warn("done")
Martin@649
   464
    if not alignments: raise Exception("there are no alignments")
Martin@641
   465
Martin@851
   466
    i1 = getSeqInfo(opts.sort1, seqNames1, seqLimits1,
Martin@851
   467
                    font, opts.fontsize, image_mode, opts.lengths1)
Martin@850
   468
    seqNames1, seqSizes1, seqLabels1, labelSizes1, margin1 = i1
Martin@846
   469
Martin@851
   470
    i2 = getSeqInfo(opts.sort2, seqNames2, seqLimits2,
Martin@851
   471
                    font, opts.fontsize, image_mode, opts.lengths2)
Martin@850
   472
    seqNames2, seqSizes2, seqLabels2, labelSizes2, margin2 = i2
Martin@641
   473
Martin@644
   474
    warn("choosing bp per pixel...")
Martin@645
   475
    pix_limit1 = opts.width  - margin1
Martin@645
   476
    pix_limit2 = opts.height - margin2
Martin@855
   477
    bpPerPix1 = get_bp_per_pix(seqSizes1, opts.border_pixels, pix_limit1)
Martin@855
   478
    bpPerPix2 = get_bp_per_pix(seqSizes2, opts.border_pixels, pix_limit2)
Martin@855
   479
    bpPerPix = max(bpPerPix1, bpPerPix2)
Martin@855
   480
    warn("bp per pixel = " + str(bpPerPix))
Martin@641
   481
Martin@855
   482
    seq_pix1, seq_starts1, width  = get_pix_info(seqSizes1, bpPerPix,
Martin@852
   483
                                                 opts.border_pixels, margin1)
Martin@855
   484
    seq_pix2, seq_starts2, height = get_pix_info(seqSizes2, bpPerPix,
Martin@852
   485
                                                 opts.border_pixels, margin2)
Martin@847
   486
    warn("width:  " + str(width))
Martin@847
   487
    warn("height: " + str(height))
Martin@839
   488
Martin@855
   489
    origins1 = dict(seqOrigins(seqNames1, seq_starts1, seqLimits1, bpPerPix))
Martin@855
   490
    origins2 = dict(seqOrigins(seqNames2, seq_starts2, seqLimits2, bpPerPix))
Martin@641
   491
Martin@644
   492
    warn("processing alignments...")
Martin@855
   493
    hits = alignmentPixels(width, height, alignments, bpPerPix,
Martin@836
   494
                           origins1, origins2)
Martin@644
   495
    warn("done")
Martin@641
   496
Martin@643
   497
    image_size = width, height
Martin@646
   498
    im = Image.new(image_mode, image_size, opts.background_color)
Martin@134
   499
Martin@845
   500
    seqLimits1 = expandedSeqDict(seqLimits1)
Martin@845
   501
    seqLimits2 = expandedSeqDict(seqLimits2)
Martin@836
   502
    origins1 = expandedSeqDict(origins1)
Martin@836
   503
    origins2 = expandedSeqDict(origins2)
Martin@845
   504
Martin@857
   505
    beds1 = itertools.chain(readBed(opts.bed1, seqLimits1),
Martin@860
   506
                            readRmsk(opts.rmsk1, seqLimits1),
Martin@860
   507
                            readGenePred(opts, opts.genePred1, seqLimits1),
Martin@857
   508
                            readGaps(opts, opts.gap1, seqLimits1))
Martin@859
   509
    b1 = bedBoxes(beds1, seqLimits1, origins1, margin2, height, True, bpPerPix)
Martin@845
   510
Martin@857
   511
    beds2 = itertools.chain(readBed(opts.bed2, seqLimits2),
Martin@860
   512
                            readRmsk(opts.rmsk2, seqLimits2),
Martin@860
   513
                            readGenePred(opts, opts.genePred2, seqLimits2),
Martin@857
   514
                            readGaps(opts, opts.gap2, seqLimits2))
Martin@859
   515
    b2 = bedBoxes(beds2, seqLimits2, origins2, margin1, width, False, bpPerPix)
Martin@857
   516
Martin@857
   517
    boxes = sorted(itertools.chain(b1, b2))
Martin@857
   518
    drawAnnotations(im, boxes)
Martin@650
   519
Martin@643
   520
    for i in range(height):
Martin@643
   521
        for j in range(width):
Martin@643
   522
            store_value = hits[i * width + j]
Martin@643
   523
            xy = j, i
Martin@643
   524
            if   store_value == 1: im.putpixel(xy, forward_color)
Martin@643
   525
            elif store_value == 2: im.putpixel(xy, reverse_color)
Martin@643
   526
            elif store_value == 3: im.putpixel(xy, overlap_color)
Martin@95
   527
Martin@643
   528
    if opts.fontsize != 0:
Martin@846
   529
        axis1 = get_axis_image(seqLabels1, labelSizes1, seq_starts1, seq_pix1,
Martin@647
   530
                               font, image_mode, opts)
Martin@846
   531
        axis2 = get_axis_image(seqLabels2, labelSizes2, seq_starts2, seq_pix2,
Martin@647
   532
                               font, image_mode, opts)
Martin@834
   533
        axis2 = axis2.transpose(Image.ROTATE_270)  # !!! bug hotspot
Martin@643
   534
        im.paste(axis1, (0, 0))
Martin@643
   535
        im.paste(axis2, (0, 0))
Martin@1
   536
Martin@643
   537
    for i in seq_starts1[1:]:
Martin@852
   538
        box = i - opts.border_pixels, margin2, i, height
Martin@852
   539
        im.paste(opts.border_color, box)
Martin@1
   540
Martin@643
   541
    for i in seq_starts2[1:]:
Martin@852
   542
        box = margin1, i - opts.border_pixels, width, i
Martin@852
   543
        im.paste(opts.border_color, box)
Martin@1
   544
Martin@643
   545
    im.save(args[1])
Martin@648
   546
Martin@648
   547
if __name__ == "__main__":
Martin@649
   548
    usage = """%prog --help
Martin@649
   549
   or: %prog [options] maf-or-tab-alignments dotplot.png
Martin@649
   550
   or: %prog [options] maf-or-tab-alignments dotplot.gif
Martin@649
   551
   or: ..."""
Martin@649
   552
    description = "Draw a dotplot of pair-wise sequence alignments in MAF or tabular format."
Martin@649
   553
    op = optparse.OptionParser(usage=usage, description=description)
Martin@866
   554
    op.add_option("-v", "--verbose", action="count",
Martin@866
   555
                  help="show progress messages & data about the plot")
Martin@651
   556
    op.add_option("-1", "--seq1", metavar="PATTERN", action="append",
Martin@840
   557
                  default=[],
Martin@651
   558
                  help="which sequences to show from the 1st genome")
Martin@651
   559
    op.add_option("-2", "--seq2", metavar="PATTERN", action="append",
Martin@840
   560
                  default=[],
Martin@651
   561
                  help="which sequences to show from the 2nd genome")
Martin@648
   562
    # Replace "width" & "height" with a single "length" option?
Martin@648
   563
    op.add_option("-x", "--width", type="int", default=1000,
Martin@648
   564
                  help="maximum width in pixels (default: %default)")
Martin@648
   565
    op.add_option("-y", "--height", type="int", default=1000,
Martin@648
   566
                  help="maximum height in pixels (default: %default)")
Martin@649
   567
    op.add_option("-c", "--forwardcolor", metavar="COLOR", default="red",
Martin@649
   568
                  help="color for forward alignments (default: %default)")
Martin@649
   569
    op.add_option("-r", "--reversecolor", metavar="COLOR", default="blue",
Martin@649
   570
                  help="color for reverse alignments (default: %default)")
Martin@851
   571
    op.add_option("--sort1", type="int", default=1, metavar="N",
Martin@851
   572
                  help="genome1 sequence order: 0=input order, 1=name order, "
Martin@851
   573
                  "2=length order (default=%default)")
Martin@851
   574
    op.add_option("--sort2", type="int", default=1, metavar="N",
Martin@851
   575
                  help="genome2 sequence order: 0=input order, 1=name order, "
Martin@851
   576
                  "2=length order (default=%default)")
Martin@839
   577
    op.add_option("--trim1", action="store_true",
Martin@839
   578
                  help="trim unaligned sequence flanks from the 1st genome")
Martin@839
   579
    op.add_option("--trim2", action="store_true",
Martin@839
   580
                  help="trim unaligned sequence flanks from the 2nd genome")
Martin@852
   581
    op.add_option("--border-pixels", metavar="INT", type="int", default=1,
Martin@852
   582
                  help="number of pixels between sequences (default=%default)")
Martin@852
   583
    op.add_option("--border-color", metavar="COLOR", default="#dcdcdc",
Martin@852
   584
                  help="color for pixels between sequences (default=%default)")
Martin@852
   585
    # xxx --margin-color?
Martin@846
   586
Martin@850
   587
    og = optparse.OptionGroup(op, "Text options")
Martin@850
   588
    og.add_option("-f", "--fontfile", metavar="FILE",
Martin@850
   589
                  help="TrueType or OpenType font file")
Martin@850
   590
    og.add_option("-s", "--fontsize", metavar="SIZE", type="int", default=11,
Martin@850
   591
                  help="TrueType or OpenType font size (default: %default)")
Martin@850
   592
    og.add_option("--lengths1", action="store_true",
Martin@850
   593
                  help="show sequence lengths for the 1st (horizontal) genome")
Martin@850
   594
    og.add_option("--lengths2", action="store_true",
Martin@850
   595
                  help="show sequence lengths for the 2nd (vertical) genome")
Martin@850
   596
    op.add_option_group(og)
Martin@850
   597
Martin@860
   598
    og = optparse.OptionGroup(op, "Annotation options")
Martin@860
   599
    og.add_option("--bed1", metavar="FILE",
Martin@860
   600
                  help="read genome1 annotations from BED file")
Martin@860
   601
    og.add_option("--bed2", metavar="FILE",
Martin@860
   602
                  help="read genome2 annotations from BED file")
Martin@860
   603
    og.add_option("--rmsk1", metavar="FILE", help="read genome1 repeats from "
Martin@860
   604
                  "RepeatMasker .out or rmsk.txt file")
Martin@860
   605
    og.add_option("--rmsk2", metavar="FILE", help="read genome2 repeats from "
Martin@860
   606
                  "RepeatMasker .out or rmsk.txt file")
Martin@860
   607
    op.add_option_group(og)
Martin@860
   608
Martin@860
   609
    og = optparse.OptionGroup(op, "Gene options")
Martin@860
   610
    og.add_option("--genePred1", metavar="FILE",
Martin@860
   611
                  help="read genome1 genes from genePred file")
Martin@860
   612
    og.add_option("--genePred2", metavar="FILE",
Martin@860
   613
                  help="read genome2 genes from genePred file")
Martin@860
   614
    og.add_option("--exon-color", metavar="COLOR", default="#dfd",
Martin@860
   615
                  help="color for exons (default=%default)")
Martin@860
   616
    og.add_option("--cds-color", metavar="COLOR", default="#bdb",
Martin@860
   617
                  help="color for protein-coding regions (default=%default)")
Martin@860
   618
    op.add_option_group(og)
Martin@860
   619
Martin@650
   620
    og = optparse.OptionGroup(op, "Unsequenced gap options")
Martin@650
   621
    og.add_option("--gap1", metavar="FILE",
Martin@650
   622
                  help="read genome1 unsequenced gaps from agp or gap file")
Martin@650
   623
    og.add_option("--gap2", metavar="FILE",
Martin@650
   624
                  help="read genome2 unsequenced gaps from agp or gap file")
Martin@650
   625
    og.add_option("--bridged-color", metavar="COLOR", default="yellow",
Martin@650
   626
                  help="color for bridged gaps (default: %default)")
Martin@650
   627
    og.add_option("--unbridged-color", metavar="COLOR", default="pink",
Martin@650
   628
                  help="color for unbridged gaps (default: %default)")
Martin@650
   629
    op.add_option_group(og)
Martin@648
   630
    (opts, args) = op.parse_args()
Martin@648
   631
    if len(args) != 2: op.error("2 arguments needed")
Martin@648
   632
Martin@648
   633
    opts.text_color = "black"
Martin@648
   634
    opts.background_color = "white"
Martin@648
   635
    opts.label_space = 5     # minimum number of pixels between axis labels
Martin@648
   636
Martin@649
   637
    try: lastDotplot(opts, args)
Martin@649
   638
    except KeyboardInterrupt: pass  # avoid silly error message
Martin@649
   639
    except Exception, e:
Martin@649
   640
        prog = os.path.basename(sys.argv[0])
Martin@649
   641
        sys.exit(prog + ": error: " + str(e))