scripts/last-dotplot
author Martin C. Frith
Thu Nov 02 22:08:14 2017 +0900 (2017-11-02)
changeset 900 5ce3e786e05a
parent 899 ccf8902bf86c
child 901 2003b209d750
permissions -rwxr-xr-x
Refactor last-dotplot
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@875
    12
import gzip
Martin@896
    13
from fnmatch import fnmatchcase
Martin@896
    14
import itertools, optparse, os, re, sys
Martin@475
    15
Martin@475
    16
# Try to make PIL/PILLOW work:
Martin@475
    17
try: from PIL import Image, ImageDraw, ImageFont, ImageColor
Martin@475
    18
except ImportError: import Image, ImageDraw, ImageFont, ImageColor
Martin@1
    19
Martin@844
    20
def myOpen(fileName):  # faster than fileinput
Martin@844
    21
    if fileName == "-":
Martin@844
    22
        return sys.stdin
Martin@875
    23
    if fileName.endswith(".gz"):
Martin@875
    24
        return gzip.open(fileName)
Martin@844
    25
    return open(fileName)
Martin@844
    26
Martin@644
    27
def warn(message):
Martin@866
    28
    if opts.verbose:
Martin@866
    29
        prog = os.path.basename(sys.argv[0])
Martin@866
    30
        sys.stderr.write(prog + ": " + message + "\n")
Martin@644
    31
Martin@840
    32
def croppedBlocks(blocks, range1, range2):
Martin@840
    33
    cropBeg1, cropEnd1 = range1
Martin@840
    34
    cropBeg2, cropEnd2 = range2
Martin@840
    35
    if blocks[0][0] < 0: cropBeg1, cropEnd1 = -cropEnd1, -cropBeg1
Martin@840
    36
    if blocks[0][1] < 0: cropBeg2, cropEnd2 = -cropEnd2, -cropBeg2
Martin@840
    37
    for beg1, beg2, size in blocks:
Martin@840
    38
        b1 = max(cropBeg1, beg1)
Martin@840
    39
        e1 = min(cropEnd1, beg1 + size)
Martin@840
    40
        if b1 >= e1: continue
Martin@840
    41
        offset = beg2 - beg1
Martin@840
    42
        b2 = max(cropBeg2, b1 + offset)
Martin@840
    43
        e2 = min(cropEnd2, e1 + offset)
Martin@840
    44
        if b2 >= e2: continue
Martin@840
    45
        yield b2 - offset, b2, e2 - b2
Martin@840
    46
Martin@482
    47
def tabBlocks(beg1, beg2, blocks):
Martin@482
    48
    '''Get the gapless blocks of an alignment, from LAST tabular format.'''
Martin@482
    49
    for i in blocks.split(","):
Martin@482
    50
        if ":" in i:
Martin@482
    51
            x, y = i.split(":")
Martin@482
    52
            beg1 += int(x)
Martin@482
    53
            beg2 += int(y)
Martin@482
    54
        else:
Martin@482
    55
            size = int(i)
Martin@482
    56
            yield beg1, beg2, size
Martin@482
    57
            beg1 += size
Martin@482
    58
            beg2 += size
Martin@272
    59
Martin@482
    60
def mafBlocks(beg1, beg2, seq1, seq2):
Martin@482
    61
    '''Get the gapless blocks of an alignment, from MAF format.'''
Martin@482
    62
    size = 0
Martin@482
    63
    for x, y in itertools.izip(seq1, seq2):
Martin@482
    64
        if x == "-":
Martin@482
    65
            if size:
Martin@482
    66
                yield beg1, beg2, size
Martin@482
    67
                beg1 += size
Martin@482
    68
                beg2 += size
Martin@482
    69
                size = 0
Martin@482
    70
            beg2 += 1
Martin@482
    71
        elif y == "-":
Martin@482
    72
            if size:
Martin@482
    73
                yield beg1, beg2, size
Martin@482
    74
                beg1 += size
Martin@482
    75
                beg2 += size
Martin@482
    76
                size = 0
Martin@482
    77
            beg1 += 1
Martin@272
    78
        else:
Martin@482
    79
            size += 1
Martin@482
    80
    if size: yield beg1, beg2, size
Martin@272
    81
Martin@482
    82
def alignmentInput(lines):
Martin@482
    83
    '''Get alignments and sequence lengths, from MAF or tabular format.'''
Martin@482
    84
    mafCount = 0
Martin@272
    85
    for line in lines:
Martin@272
    86
        w = line.split()
Martin@272
    87
        if line[0].isdigit():  # tabular format
Martin@482
    88
            chr1, beg1, seqlen1 = w[1], int(w[2]), int(w[5])
Martin@482
    89
            if w[4] == "-": beg1 -= seqlen1
Martin@482
    90
            chr2, beg2, seqlen2 = w[6], int(w[7]), int(w[10])
Martin@482
    91
            if w[9] == "-": beg2 -= seqlen2
Martin@847
    92
            blocks = tabBlocks(beg1, beg2, w[11])
Martin@482
    93
            yield chr1, seqlen1, chr2, seqlen2, blocks
Martin@272
    94
        elif line[0] == "s":  # MAF format
Martin@482
    95
            if mafCount == 0:
Martin@482
    96
                chr1, beg1, seqlen1, seq1 = w[1], int(w[2]), int(w[5]), w[6]
Martin@482
    97
                if w[4] == "-": beg1 -= seqlen1
Martin@482
    98
                mafCount = 1
Martin@482
    99
            else:
Martin@482
   100
                chr2, beg2, seqlen2, seq2 = w[1], int(w[2]), int(w[5]), w[6]
Martin@482
   101
                if w[4] == "-": beg2 -= seqlen2
Martin@847
   102
                blocks = mafBlocks(beg1, beg2, seq1, seq2)
Martin@482
   103
                yield chr1, seqlen1, chr2, seqlen2, blocks
Martin@482
   104
                mafCount = 0
Martin@272
   105
Martin@897
   106
def seqRequestFromText(text):
Martin@840
   107
    if ":" in text:
Martin@840
   108
        pattern, interval = text.rsplit(":", 1)
Martin@840
   109
        if "-" in interval:
Martin@840
   110
            beg, end = interval.rsplit("-", 1)
Martin@840
   111
            return pattern, int(beg), int(end)  # beg may be negative
Martin@840
   112
    return text, 0, sys.maxsize
Martin@840
   113
Martin@897
   114
def rangeFromSeqName(seqRequests, name, seqLen):
Martin@897
   115
    if not seqRequests: return 0, seqLen
Martin@651
   116
    base = name.split(".")[-1]  # allow for names like hg19.chr7
Martin@897
   117
    for pat, beg, end in seqRequests:
Martin@896
   118
        if fnmatchcase(name, pat) or fnmatchcase(base, pat):
Martin@840
   119
            return max(beg, 0), min(end, seqLen)
Martin@844
   120
    return None
Martin@651
   121
Martin@851
   122
def updateSeqs(isTrim, seqNames, seqLimits, seqName, seqRange, blocks, index):
Martin@851
   123
    if seqName not in seqLimits:
Martin@851
   124
        seqNames.append(seqName)
Martin@839
   125
    if isTrim:
Martin@839
   126
        beg = blocks[0][index]
Martin@839
   127
        end = blocks[-1][index] + blocks[-1][2]
Martin@839
   128
        if beg < 0: beg, end = -end, -beg
Martin@839
   129
        if seqName in seqLimits:
Martin@839
   130
            b, e = seqLimits[seqName]
Martin@839
   131
            seqLimits[seqName] = min(b, beg), max(e, end)
Martin@839
   132
        else:
Martin@839
   133
            seqLimits[seqName] = beg, end
Martin@839
   134
    else:
Martin@840
   135
        seqLimits[seqName] = seqRange
Martin@839
   136
Martin@651
   137
def readAlignments(fileName, opts):
Martin@839
   138
    '''Get alignments and sequence limits, from MAF or tabular format.'''
Martin@897
   139
    seqRequests1 = map(seqRequestFromText, opts.seq1)
Martin@897
   140
    seqRequests2 = map(seqRequestFromText, opts.seq2)
Martin@840
   141
Martin@482
   142
    alignments = []
Martin@851
   143
    seqNames1 = []
Martin@851
   144
    seqNames2 = []
Martin@839
   145
    seqLimits1 = {}
Martin@839
   146
    seqLimits2 = {}
Martin@844
   147
    lines = myOpen(fileName)
Martin@838
   148
    for seqName1, seqLen1, seqName2, seqLen2, blocks in alignmentInput(lines):
Martin@897
   149
        range1 = rangeFromSeqName(seqRequests1, seqName1, seqLen1)
Martin@844
   150
        if not range1: continue
Martin@897
   151
        range2 = rangeFromSeqName(seqRequests2, seqName2, seqLen2)
Martin@844
   152
        if not range2: continue
Martin@847
   153
        b = list(croppedBlocks(list(blocks), range1, range2))
Martin@840
   154
        if not b: continue
Martin@840
   155
        aln = seqName1, seqName2, b
Martin@482
   156
        alignments.append(aln)
Martin@851
   157
        updateSeqs(opts.trim1, seqNames1, seqLimits1, seqName1, range1, b, 0)
Martin@851
   158
        updateSeqs(opts.trim2, seqNames2, seqLimits2, seqName2, range2, b, 1)
Martin@851
   159
    return alignments, seqNames1, seqNames2, seqLimits1, seqLimits2
Martin@1
   160
Martin@1
   161
def natural_sort_key(my_string):
Martin@1
   162
    '''Return a sort key for "natural" ordering, e.g. chr9 < chr10.'''
Martin@1
   163
    parts = re.split(r'(\d+)', my_string)
Martin@1
   164
    parts[1::2] = map(int, parts[1::2])
Martin@1
   165
    return parts
Martin@1
   166
Martin@898
   167
def prettyNum(n):
Martin@898
   168
    t = str(n)
Martin@898
   169
    groups = []
Martin@898
   170
    while t:
Martin@898
   171
        groups.append(t[-3:])
Martin@898
   172
        t = t[:-3]
Martin@898
   173
    return " ".join(reversed(groups))
Martin@898
   174
Martin@846
   175
def sizeText(size):
Martin@846
   176
    suffixes = "bp", "kb", "Mb", "Gb"
Martin@846
   177
    for i, x in enumerate(suffixes):
Martin@846
   178
        j = 10 ** (i * 3)
Martin@846
   179
        if size < j * 10:
Martin@846
   180
            return "%.2g" % (1.0 * size / j) + x
Martin@846
   181
        if size < j * 1000 or i == len(suffixes) - 1:
Martin@846
   182
            return "%.0f" % (1.0 * size / j) + x
Martin@846
   183
Martin@898
   184
def labelText(seqRange, labelOpt):
Martin@898
   185
    seqName, beg, end = seqRange
Martin@898
   186
    if labelOpt == 1:
Martin@898
   187
        return seqName + ": " + sizeText(end - beg)
Martin@898
   188
    if labelOpt == 2:
Martin@898
   189
        return seqName + ": " + prettyNum(beg) + ": " + sizeText(end - beg)
Martin@898
   190
    if labelOpt == 3:
Martin@898
   191
        return seqName + ": " + prettyNum(beg) + " - " + prettyNum(end)
Martin@898
   192
    return seqName
Martin@846
   193
Martin@899
   194
def rangeLabels(seqRanges, labelOpt, font, fontsize, image_mode, textRot):
Martin@899
   195
    if fontsize:
Martin@899
   196
        image_size = 1, 1
Martin@899
   197
        im = Image.new(image_mode, image_size)
Martin@899
   198
        draw = ImageDraw.Draw(im)
Martin@899
   199
    x = y = 0
Martin@899
   200
    for r in seqRanges:
Martin@899
   201
        text = labelText(r, labelOpt)
Martin@899
   202
        if fontsize:
Martin@899
   203
            x, y = draw.textsize(text, font=font)
Martin@899
   204
            if textRot:
Martin@899
   205
                x, y = y, x
Martin@899
   206
        yield text, x, y
Martin@899
   207
Martin@851
   208
def getSeqInfo(sortOpt, seqNames, seqLimits,
Martin@898
   209
               font, fontsize, image_mode, labelOpt, textRot):
Martin@1
   210
    '''Return miscellaneous information about the sequences.'''
Martin@851
   211
    if sortOpt == 1:
Martin@851
   212
        seqNames.sort(key=natural_sort_key)
Martin@850
   213
    seqSizes = [seqLimits[i][1] - seqLimits[i][0] for i in seqNames]
Martin@866
   214
    for i in seqNames:
Martin@866
   215
        r = seqLimits[i]
Martin@866
   216
        out = i, str(r[0]), str(r[1])
Martin@866
   217
        warn("\t".join(out))
Martin@866
   218
    warn("")
Martin@851
   219
    if sortOpt == 2:
Martin@851
   220
        seqRecords = sorted(zip(seqSizes, seqNames), reverse=True)
Martin@851
   221
        seqSizes = [i[0] for i in seqRecords]
Martin@851
   222
        seqNames = [i[1] for i in seqRecords]
Martin@898
   223
    seqRanges = [(i, seqLimits[i][0], seqLimits[i][1]) for i in seqNames]
Martin@899
   224
    labelData = list(rangeLabels(seqRanges, labelOpt,
Martin@899
   225
                                 font, fontsize, image_mode, textRot))
Martin@899
   226
    margin = max(i[2] for i in labelData)
Martin@878
   227
    # xxx the margin may be too big, because some labels may get omitted
Martin@899
   228
    return seqNames, seqSizes, labelData, margin
Martin@1
   229
Martin@1
   230
def div_ceil(x, y):
Martin@1
   231
    '''Return x / y rounded up.'''
Martin@1
   232
    q, r = divmod(x, y)
Martin@1
   233
    return q + (r != 0)
Martin@1
   234
Martin@896
   235
def get_bp_per_pix(rangeSizes, pixTweenRanges, maxPixels):
Martin@1
   236
    '''Get the minimum bp-per-pixel that fits in the size limit.'''
Martin@896
   237
    numOfRanges = len(rangeSizes)
Martin@896
   238
    maxPixelsInRanges = maxPixels - pixTweenRanges * (numOfRanges - 1)
Martin@896
   239
    if maxPixelsInRanges < numOfRanges:
Martin@649
   240
        raise Exception("can't fit the image: too many sequences?")
Martin@896
   241
    negLimit = -maxPixelsInRanges
Martin@896
   242
    negBpPerPix = sum(rangeSizes) // negLimit
Martin@863
   243
    while True:
Martin@896
   244
        if sum(i // negBpPerPix for i in rangeSizes) >= negLimit:
Martin@863
   245
            return -negBpPerPix
Martin@863
   246
        negBpPerPix -= 1
Martin@1
   247
Martin@900
   248
def getRangePixBegs(rangePixLens, pixTweenRanges, margin):
Martin@900
   249
    '''Get the start pixel for each range.'''
Martin@900
   250
    rangePixBegs = []
Martin@896
   251
    pix_tot = margin - pixTweenRanges
Martin@900
   252
    for i in rangePixLens:
Martin@896
   253
        pix_tot += pixTweenRanges
Martin@900
   254
        rangePixBegs.append(pix_tot)
Martin@1
   255
        pix_tot += i
Martin@900
   256
    return rangePixBegs
Martin@1
   257
Martin@896
   258
def pixelData(rangeSizes, bp_per_pix, pixTweenRanges, margin):
Martin@900
   259
    '''Return pixel information about the ranges.'''
Martin@900
   260
    rangePixLens = [div_ceil(i, bp_per_pix) for i in rangeSizes]
Martin@900
   261
    rangePixBegs = getRangePixBegs(rangePixLens, pixTweenRanges, margin)
Martin@900
   262
    tot_pix = rangePixBegs[-1] + rangePixLens[-1]
Martin@900
   263
    return rangePixBegs, rangePixLens, tot_pix
Martin@1
   264
Martin@835
   265
def drawLineForward(hits, width, bp_per_pix, beg1, beg2, size):
Martin@639
   266
    while True:
Martin@639
   267
        q1, r1 = divmod(beg1, bp_per_pix)
Martin@639
   268
        q2, r2 = divmod(beg2, bp_per_pix)
Martin@835
   269
        hits[q2 * width + q1] |= 1
Martin@639
   270
        next_pix = min(bp_per_pix - r1, bp_per_pix - r2)
Martin@639
   271
        if next_pix >= size: break
Martin@639
   272
        beg1 += next_pix
Martin@639
   273
        beg2 += next_pix
Martin@639
   274
        size -= next_pix
Martin@639
   275
Martin@835
   276
def drawLineReverse(hits, width, bp_per_pix, beg1, beg2, size):
Martin@639
   277
    beg2 = -1 - beg2
Martin@639
   278
    while True:
Martin@639
   279
        q1, r1 = divmod(beg1, bp_per_pix)
Martin@639
   280
        q2, r2 = divmod(beg2, bp_per_pix)
Martin@835
   281
        hits[q2 * width + q1] |= 2
Martin@639
   282
        next_pix = min(bp_per_pix - r1, r2 + 1)
Martin@639
   283
        if next_pix >= size: break
Martin@639
   284
        beg1 += next_pix
Martin@639
   285
        beg2 -= next_pix
Martin@639
   286
        size -= next_pix
Martin@639
   287
Martin@836
   288
def alignmentPixels(width, height, alignments, bp_per_pix, origins1, origins2):
Martin@640
   289
    hits = [0] * (width * height)  # the image data
Martin@640
   290
    for seq1, seq2, blocks in alignments:
Martin@836
   291
        ori1 = origins1[seq1]
Martin@836
   292
        ori2 = origins2[seq2]
Martin@640
   293
        for beg1, beg2, size in blocks:
Martin@640
   294
            if beg1 < 0:
Martin@640
   295
                beg1 = -(beg1 + size)
Martin@640
   296
                beg2 = -(beg2 + size)
Martin@640
   297
            if beg2 >= 0:
Martin@835
   298
                drawLineForward(hits, width, bp_per_pix,
Martin@835
   299
                                beg1 + ori1, beg2 + ori2, size)
Martin@640
   300
            else:
Martin@835
   301
                drawLineReverse(hits, width, bp_per_pix,
Martin@835
   302
                                beg1 + ori1, beg2 - ori2, size)
Martin@640
   303
    return hits
Martin@1
   304
Martin@650
   305
def expandedSeqDict(seqDict):
Martin@650
   306
    '''Allow lookup by short sequence names, e.g. chr7 as well as hg19.chr7.'''
Martin@861
   307
    newDict = seqDict.copy()
Martin@650
   308
    for name, x in seqDict.items():
Martin@861
   309
        if "." in name:
Martin@861
   310
            base = name.split(".")[-1]
Martin@861
   311
            if base in newDict:  # an ambiguous case was found:
Martin@861
   312
                return seqDict   # so give up completely
Martin@861
   313
            newDict[base] = x
Martin@650
   314
    return newDict
Martin@650
   315
Martin@845
   316
def readBed(fileName, seqLimits):
Martin@845
   317
    if not fileName: return
Martin@845
   318
    for line in myOpen(fileName):
Martin@845
   319
        w = line.split()
Martin@847
   320
        if not w: continue
Martin@845
   321
        seqName = w[0]
Martin@845
   322
        if seqName not in seqLimits: continue
Martin@845
   323
        beg = int(w[1])
Martin@845
   324
        end = int(w[2])
Martin@857
   325
        layer = 900
Martin@895
   326
        color = "#fbf"
Martin@858
   327
        if len(w) > 4:
Martin@858
   328
            if w[4] != ".":
Martin@858
   329
                layer = float(w[4])
Martin@858
   330
            if len(w) > 5:
Martin@858
   331
                if len(w) > 8 and w[8].count(",") == 2:
Martin@858
   332
                    color = "rgb(" + w[8] + ")"
Martin@858
   333
                elif w[5] == "+":
Martin@895
   334
                    color = "#ffe8e8"
Martin@858
   335
                elif w[5] == "-":
Martin@895
   336
                    color = "#e8e8ff"
Martin@859
   337
        yield layer, color, seqName, beg, end
Martin@845
   338
Martin@860
   339
def commaSeparatedInts(text):
Martin@860
   340
    return map(int, text.rstrip(",").split(","))
Martin@860
   341
Martin@860
   342
def readGenePred(opts, fileName, seqLimits):
Martin@860
   343
    if not fileName: return
Martin@860
   344
    for line in myOpen(fileName):
Martin@860
   345
        fields = line.split()
Martin@860
   346
        if not fields: continue
Martin@860
   347
        if fields[2] not in "+-": fields = fields[1:]
Martin@860
   348
        seqName = fields[1]
Martin@860
   349
        if seqName not in seqLimits: continue
Martin@860
   350
        #strand = fields[2]
Martin@860
   351
        cdsBeg = int(fields[5])
Martin@860
   352
        cdsEnd = int(fields[6])
Martin@860
   353
        exonBegs = commaSeparatedInts(fields[8])
Martin@860
   354
        exonEnds = commaSeparatedInts(fields[9])
Martin@860
   355
        for beg, end in zip(exonBegs, exonEnds):
Martin@860
   356
            yield 300, opts.exon_color, seqName, beg, end
Martin@860
   357
            b = max(beg, cdsBeg)
Martin@860
   358
            e = min(end, cdsEnd)
Martin@860
   359
            if b < e: yield 400, opts.cds_color, seqName, b, e
Martin@860
   360
Martin@860
   361
def readRmsk(fileName, seqLimits):
Martin@860
   362
    if not fileName: return
Martin@860
   363
    for line in myOpen(fileName):
Martin@860
   364
        fields = line.split()
Martin@860
   365
        if len(fields) == 17:  # rmsk.txt
Martin@860
   366
            seqName = fields[5]
Martin@860
   367
            if seqName not in seqLimits: continue  # do this ASAP for speed
Martin@860
   368
            beg = int(fields[6])
Martin@860
   369
            end = int(fields[7])
Martin@860
   370
            strand = fields[9]
Martin@860
   371
            repeatClass = fields[11]
Martin@860
   372
        elif len(fields) == 15:  # .out
Martin@860
   373
            seqName = fields[4]
Martin@860
   374
            if seqName not in seqLimits: continue
Martin@860
   375
            beg = int(fields[5]) - 1
Martin@860
   376
            end = int(fields[6])
Martin@860
   377
            strand = fields[8]
Martin@860
   378
            repeatClass = fields[10]
Martin@860
   379
        else:
Martin@860
   380
            continue
Martin@860
   381
        if repeatClass in ("Low_complexity", "Simple_repeat"):
Martin@895
   382
            yield 200, "#fbf", seqName, beg, end
Martin@860
   383
        elif strand == "+":
Martin@895
   384
            yield 100, "#ffe8e8", seqName, beg, end
Martin@860
   385
        else:
Martin@895
   386
            yield 100, "#e8e8ff", seqName, beg, end
Martin@860
   387
Martin@650
   388
def isExtraFirstGapField(fields):
Martin@650
   389
    return fields[4].isdigit()
Martin@650
   390
Martin@857
   391
def readGaps(opts, fileName, seqLimits):
Martin@650
   392
    '''Read locations of unsequenced gaps, from an agp or gap file.'''
Martin@650
   393
    if not fileName: return
Martin@844
   394
    for line in myOpen(fileName):
Martin@650
   395
        w = line.split()
Martin@650
   396
        if not w or w[0][0] == "#": continue
Martin@650
   397
        if isExtraFirstGapField(w): w = w[1:]
Martin@650
   398
        if w[4] not in "NU": continue
Martin@650
   399
        seqName = w[0]
Martin@839
   400
        if seqName not in seqLimits: continue
Martin@650
   401
        end = int(w[2])
Martin@650
   402
        beg = end - int(w[5])  # zero-based coordinate
Martin@857
   403
        if w[7] == "yes":
Martin@859
   404
            yield 3000, opts.bridged_color, seqName, beg, end
Martin@857
   405
        else:
Martin@859
   406
            yield 2000, opts.unbridged_color, seqName, beg, end
Martin@650
   407
Martin@859
   408
def bedBoxes(beds, seqLimits, origins, margin, edge, isTop, bpPerPix):
Martin@897
   409
    for layer, color, seqName, bedBeg, bedEnd in beds:
Martin@859
   410
        cropBeg, cropEnd = seqLimits[seqName]
Martin@897
   411
        beg = max(bedBeg, cropBeg)
Martin@897
   412
        end = min(bedEnd, cropEnd)
Martin@859
   413
        if beg >= end: continue
Martin@845
   414
        ori = origins[seqName]
Martin@857
   415
        if layer <= 1000:
Martin@857
   416
            # include partly-covered pixels
Martin@857
   417
            b = (ori + beg) // bpPerPix
Martin@857
   418
            e = div_ceil(ori + end, bpPerPix)
Martin@857
   419
        else:
Martin@857
   420
            # exclude partly-covered pixels
Martin@857
   421
            b = div_ceil(ori + beg, bpPerPix)
Martin@857
   422
            e = (ori + end) // bpPerPix
Martin@857
   423
            if e <= b: continue
Martin@857
   424
        if isTop:
Martin@857
   425
            box = b, margin, e, edge
Martin@857
   426
        else:
Martin@857
   427
            box = margin, b, edge, e
Martin@857
   428
        yield layer, color, box
Martin@845
   429
Martin@857
   430
def drawAnnotations(im, boxes):
Martin@857
   431
    # xxx use partial transparency for different-color overlaps?
Martin@857
   432
    for layer, color, box in boxes:
Martin@650
   433
        im.paste(color, box)
Martin@650
   434
Martin@899
   435
def make_label(labelData, range_start, range_size):
Martin@1
   436
    '''Return an axis label with endpoint & sort-order information.'''
Martin@899
   437
    text, text_width, text_height = labelData
Martin@1
   438
    label_start = range_start + (range_size - text_width) // 2
Martin@1
   439
    label_end   = label_start + text_width
Martin@1
   440
    sort_key    = text_width - range_size
Martin@878
   441
    return sort_key, label_start, label_end, text, text_height
Martin@1
   442
Martin@897
   443
def nonoverlappingLabels(labels, minPixTweenLabels):
Martin@1
   444
    '''Get a subset of non-overlapping axis labels, greedily.'''
Martin@897
   445
    out = []
Martin@1
   446
    for i in labels:
Martin@897
   447
        beg = i[1] - minPixTweenLabels
Martin@897
   448
        end = i[2] + minPixTweenLabels
Martin@897
   449
        if all(j[2] <= beg or j[1] >= end for j in out):
Martin@897
   450
            out.append(i)
Martin@897
   451
    return out
Martin@1
   452
Martin@900
   453
def axisImage(labelData, rangePixBegs, rangePixLens, textRot,
Martin@895
   454
              textAln, font, image_mode, opts):
Martin@1
   455
    '''Make an image of axis labels.'''
Martin@900
   456
    beg = rangePixBegs[0]
Martin@900
   457
    end = rangePixBegs[-1] + rangePixLens[-1]
Martin@899
   458
    margin = max(i[2] for i in labelData)
Martin@900
   459
    labels = map(make_label, labelData, rangePixBegs, rangePixLens)
Martin@897
   460
    labels = [i for i in labels if i[1] >= beg and i[2] <= end]
Martin@1
   461
    labels.sort()
Martin@878
   462
    minPixTweenLabels = 0 if textRot else opts.label_space
Martin@897
   463
    labels = nonoverlappingLabels(labels, minPixTweenLabels)
Martin@897
   464
    image_size = (margin, end) if textRot else (end, margin)
Martin@895
   465
    im = Image.new(image_mode, image_size, opts.margin_color)
Martin@1
   466
    draw = ImageDraw.Draw(im)
Martin@1
   467
    for i in labels:
Martin@878
   468
        base = margin - i[4] if textAln else 0
Martin@878
   469
        position = (base, i[1]) if textRot else (i[1], base)
Martin@646
   470
        draw.text(position, i[3], font=font, fill=opts.text_color)
Martin@1
   471
    return im
Martin@1
   472
Martin@900
   473
def seqOrigins(seqNames, rangePixBegs, seqLimits, bp_per_pix):
Martin@900
   474
    for i, j in zip(seqNames, rangePixBegs):
Martin@839
   475
        yield i, bp_per_pix * j - seqLimits[i][0]
Martin@836
   476
Martin@648
   477
def lastDotplot(opts, args):
Martin@643
   478
    if opts.fontfile:  font = ImageFont.truetype(opts.fontfile, opts.fontsize)
Martin@643
   479
    else:              font = ImageFont.load_default()
Martin@641
   480
Martin@643
   481
    image_mode = 'RGB'
Martin@643
   482
    forward_color = ImageColor.getcolor(opts.forwardcolor, image_mode)
Martin@643
   483
    reverse_color = ImageColor.getcolor(opts.reversecolor, image_mode)
Martin@643
   484
    zipped_colors = zip(forward_color, reverse_color)
Martin@643
   485
    overlap_color = tuple([(i + j) // 2 for i, j in zipped_colors])
Martin@641
   486
Martin@644
   487
    warn("reading alignments...")
Martin@851
   488
    alignmentInfo = readAlignments(args[0], opts)
Martin@851
   489
    alignments, seqNames1, seqNames2, seqLimits1, seqLimits2 = alignmentInfo
Martin@644
   490
    warn("done")
Martin@649
   491
    if not alignments: raise Exception("there are no alignments")
Martin@641
   492
Martin@878
   493
    textRot1 = "vertical".startswith(opts.rot1)
Martin@851
   494
    i1 = getSeqInfo(opts.sort1, seqNames1, seqLimits1,
Martin@898
   495
                    font, opts.fontsize, image_mode, opts.labels1, textRot1)
Martin@899
   496
    seqNames1, rangeSizes1, labelData1, tMargin = i1
Martin@846
   497
Martin@878
   498
    textRot2 = "horizontal".startswith(opts.rot2)
Martin@851
   499
    i2 = getSeqInfo(opts.sort2, seqNames2, seqLimits2,
Martin@898
   500
                    font, opts.fontsize, image_mode, opts.labels2, textRot2)
Martin@899
   501
    seqNames2, rangeSizes2, labelData2, lMargin = i2
Martin@641
   502
Martin@644
   503
    warn("choosing bp per pixel...")
Martin@896
   504
    maxPixels1 = opts.width  - lMargin
Martin@896
   505
    maxPixels2 = opts.height - tMargin
Martin@897
   506
    bpPerPix1 = get_bp_per_pix(rangeSizes1, opts.border_pixels, maxPixels1)
Martin@897
   507
    bpPerPix2 = get_bp_per_pix(rangeSizes2, opts.border_pixels, maxPixels2)
Martin@855
   508
    bpPerPix = max(bpPerPix1, bpPerPix2)
Martin@855
   509
    warn("bp per pixel = " + str(bpPerPix))
Martin@641
   510
Martin@900
   511
    p1 = pixelData(rangeSizes1, bpPerPix, opts.border_pixels, lMargin)
Martin@900
   512
    rangePixBegs1, rangePixLens1, width = p1
Martin@900
   513
Martin@900
   514
    p2 = pixelData(rangeSizes2, bpPerPix, opts.border_pixels, tMargin)
Martin@900
   515
    rangePixBegs2, rangePixLens2, height = p2
Martin@900
   516
Martin@847
   517
    warn("width:  " + str(width))
Martin@847
   518
    warn("height: " + str(height))
Martin@839
   519
Martin@900
   520
    origins1 = dict(seqOrigins(seqNames1, rangePixBegs1, seqLimits1, bpPerPix))
Martin@900
   521
    origins2 = dict(seqOrigins(seqNames2, rangePixBegs2, seqLimits2, bpPerPix))
Martin@641
   522
Martin@644
   523
    warn("processing alignments...")
Martin@855
   524
    hits = alignmentPixels(width, height, alignments, bpPerPix,
Martin@836
   525
                           origins1, origins2)
Martin@644
   526
    warn("done")
Martin@641
   527
Martin@643
   528
    image_size = width, height
Martin@646
   529
    im = Image.new(image_mode, image_size, opts.background_color)
Martin@134
   530
Martin@845
   531
    seqLimits1 = expandedSeqDict(seqLimits1)
Martin@845
   532
    seqLimits2 = expandedSeqDict(seqLimits2)
Martin@836
   533
    origins1 = expandedSeqDict(origins1)
Martin@836
   534
    origins2 = expandedSeqDict(origins2)
Martin@845
   535
Martin@857
   536
    beds1 = itertools.chain(readBed(opts.bed1, seqLimits1),
Martin@860
   537
                            readRmsk(opts.rmsk1, seqLimits1),
Martin@860
   538
                            readGenePred(opts, opts.genePred1, seqLimits1),
Martin@857
   539
                            readGaps(opts, opts.gap1, seqLimits1))
Martin@877
   540
    b1 = bedBoxes(beds1, seqLimits1, origins1, tMargin, height, True, bpPerPix)
Martin@845
   541
Martin@857
   542
    beds2 = itertools.chain(readBed(opts.bed2, seqLimits2),
Martin@860
   543
                            readRmsk(opts.rmsk2, seqLimits2),
Martin@860
   544
                            readGenePred(opts, opts.genePred2, seqLimits2),
Martin@857
   545
                            readGaps(opts, opts.gap2, seqLimits2))
Martin@877
   546
    b2 = bedBoxes(beds2, seqLimits2, origins2, lMargin, width, False, bpPerPix)
Martin@857
   547
Martin@857
   548
    boxes = sorted(itertools.chain(b1, b2))
Martin@857
   549
    drawAnnotations(im, boxes)
Martin@650
   550
Martin@643
   551
    for i in range(height):
Martin@643
   552
        for j in range(width):
Martin@643
   553
            store_value = hits[i * width + j]
Martin@643
   554
            xy = j, i
Martin@643
   555
            if   store_value == 1: im.putpixel(xy, forward_color)
Martin@643
   556
            elif store_value == 2: im.putpixel(xy, reverse_color)
Martin@643
   557
            elif store_value == 3: im.putpixel(xy, overlap_color)
Martin@95
   558
Martin@643
   559
    if opts.fontsize != 0:
Martin@900
   560
        axis1 = axisImage(labelData1, rangePixBegs1, rangePixLens1,
Martin@895
   561
                          textRot1, False, font, image_mode, opts)
Martin@878
   562
        if textRot1:
Martin@878
   563
            axis1 = axis1.transpose(Image.ROTATE_90)
Martin@900
   564
        axis2 = axisImage(labelData2, rangePixBegs2, rangePixLens2,
Martin@895
   565
                          textRot2, textRot2, font, image_mode, opts)
Martin@878
   566
        if not textRot2:
Martin@878
   567
            axis2 = axis2.transpose(Image.ROTATE_270)
Martin@643
   568
        im.paste(axis1, (0, 0))
Martin@643
   569
        im.paste(axis2, (0, 0))
Martin@1
   570
Martin@900
   571
    for i in rangePixBegs1[1:]:
Martin@877
   572
        box = i - opts.border_pixels, tMargin, i, height
Martin@852
   573
        im.paste(opts.border_color, box)
Martin@1
   574
Martin@900
   575
    for i in rangePixBegs2[1:]:
Martin@877
   576
        box = lMargin, i - opts.border_pixels, width, i
Martin@852
   577
        im.paste(opts.border_color, box)
Martin@1
   578
Martin@643
   579
    im.save(args[1])
Martin@648
   580
Martin@648
   581
if __name__ == "__main__":
Martin@649
   582
    usage = """%prog --help
Martin@649
   583
   or: %prog [options] maf-or-tab-alignments dotplot.png
Martin@649
   584
   or: %prog [options] maf-or-tab-alignments dotplot.gif
Martin@649
   585
   or: ..."""
Martin@649
   586
    description = "Draw a dotplot of pair-wise sequence alignments in MAF or tabular format."
Martin@649
   587
    op = optparse.OptionParser(usage=usage, description=description)
Martin@866
   588
    op.add_option("-v", "--verbose", action="count",
Martin@866
   589
                  help="show progress messages & data about the plot")
Martin@651
   590
    op.add_option("-1", "--seq1", metavar="PATTERN", action="append",
Martin@840
   591
                  default=[],
Martin@651
   592
                  help="which sequences to show from the 1st genome")
Martin@651
   593
    op.add_option("-2", "--seq2", metavar="PATTERN", action="append",
Martin@840
   594
                  default=[],
Martin@651
   595
                  help="which sequences to show from the 2nd genome")
Martin@648
   596
    # Replace "width" & "height" with a single "length" option?
Martin@648
   597
    op.add_option("-x", "--width", type="int", default=1000,
Martin@648
   598
                  help="maximum width in pixels (default: %default)")
Martin@648
   599
    op.add_option("-y", "--height", type="int", default=1000,
Martin@648
   600
                  help="maximum height in pixels (default: %default)")
Martin@649
   601
    op.add_option("-c", "--forwardcolor", metavar="COLOR", default="red",
Martin@649
   602
                  help="color for forward alignments (default: %default)")
Martin@649
   603
    op.add_option("-r", "--reversecolor", metavar="COLOR", default="blue",
Martin@649
   604
                  help="color for reverse alignments (default: %default)")
Martin@851
   605
    op.add_option("--sort1", type="int", default=1, metavar="N",
Martin@851
   606
                  help="genome1 sequence order: 0=input order, 1=name order, "
Martin@851
   607
                  "2=length order (default=%default)")
Martin@851
   608
    op.add_option("--sort2", type="int", default=1, metavar="N",
Martin@851
   609
                  help="genome2 sequence order: 0=input order, 1=name order, "
Martin@851
   610
                  "2=length order (default=%default)")
Martin@839
   611
    op.add_option("--trim1", action="store_true",
Martin@839
   612
                  help="trim unaligned sequence flanks from the 1st genome")
Martin@839
   613
    op.add_option("--trim2", action="store_true",
Martin@839
   614
                  help="trim unaligned sequence flanks from the 2nd genome")
Martin@852
   615
    op.add_option("--border-pixels", metavar="INT", type="int", default=1,
Martin@852
   616
                  help="number of pixels between sequences (default=%default)")
Martin@895
   617
    op.add_option("--border-color", metavar="COLOR", default="black",
Martin@852
   618
                  help="color for pixels between sequences (default=%default)")
Martin@895
   619
    op.add_option("--margin-color", metavar="COLOR", default="#dcdcdc",
Martin@895
   620
                  help="margin color")
Martin@846
   621
Martin@850
   622
    og = optparse.OptionGroup(op, "Text options")
Martin@850
   623
    og.add_option("-f", "--fontfile", metavar="FILE",
Martin@850
   624
                  help="TrueType or OpenType font file")
Martin@850
   625
    og.add_option("-s", "--fontsize", metavar="SIZE", type="int", default=11,
Martin@850
   626
                  help="TrueType or OpenType font size (default: %default)")
Martin@898
   627
    og.add_option("--labels1", type="int", default=0, metavar="N", help=
Martin@898
   628
                  "genome1 labels: 0=name, 1=name:length, "
Martin@898
   629
                  "2=name:start:length, 3=name:start-end (default=%default)")
Martin@898
   630
    og.add_option("--labels2", type="int", default=0, metavar="N", help=
Martin@898
   631
                  "genome2 labels: 0=name, 1=name:length, "
Martin@898
   632
                  "2=name:start:length, 3=name:start-end (default=%default)")
Martin@878
   633
    og.add_option("--rot1", metavar="ROT", default="h",
Martin@878
   634
                  help="text rotation for the 1st genome (default=%default)")
Martin@878
   635
    og.add_option("--rot2", metavar="ROT", default="v",
Martin@878
   636
                  help="text rotation for the 2nd genome (default=%default)")
Martin@850
   637
    op.add_option_group(og)
Martin@850
   638
Martin@860
   639
    og = optparse.OptionGroup(op, "Annotation options")
Martin@860
   640
    og.add_option("--bed1", metavar="FILE",
Martin@860
   641
                  help="read genome1 annotations from BED file")
Martin@860
   642
    og.add_option("--bed2", metavar="FILE",
Martin@860
   643
                  help="read genome2 annotations from BED file")
Martin@860
   644
    og.add_option("--rmsk1", metavar="FILE", help="read genome1 repeats from "
Martin@860
   645
                  "RepeatMasker .out or rmsk.txt file")
Martin@860
   646
    og.add_option("--rmsk2", metavar="FILE", help="read genome2 repeats from "
Martin@860
   647
                  "RepeatMasker .out or rmsk.txt file")
Martin@860
   648
    op.add_option_group(og)
Martin@860
   649
Martin@860
   650
    og = optparse.OptionGroup(op, "Gene options")
Martin@860
   651
    og.add_option("--genePred1", metavar="FILE",
Martin@860
   652
                  help="read genome1 genes from genePred file")
Martin@860
   653
    og.add_option("--genePred2", metavar="FILE",
Martin@860
   654
                  help="read genome2 genes from genePred file")
Martin@895
   655
    og.add_option("--exon-color", metavar="COLOR", default="PaleGreen",
Martin@860
   656
                  help="color for exons (default=%default)")
Martin@895
   657
    og.add_option("--cds-color", metavar="COLOR", default="LimeGreen",
Martin@860
   658
                  help="color for protein-coding regions (default=%default)")
Martin@860
   659
    op.add_option_group(og)
Martin@860
   660
Martin@650
   661
    og = optparse.OptionGroup(op, "Unsequenced gap options")
Martin@650
   662
    og.add_option("--gap1", metavar="FILE",
Martin@650
   663
                  help="read genome1 unsequenced gaps from agp or gap file")
Martin@650
   664
    og.add_option("--gap2", metavar="FILE",
Martin@650
   665
                  help="read genome2 unsequenced gaps from agp or gap file")
Martin@650
   666
    og.add_option("--bridged-color", metavar="COLOR", default="yellow",
Martin@650
   667
                  help="color for bridged gaps (default: %default)")
Martin@895
   668
    og.add_option("--unbridged-color", metavar="COLOR", default="orange",
Martin@650
   669
                  help="color for unbridged gaps (default: %default)")
Martin@650
   670
    op.add_option_group(og)
Martin@648
   671
    (opts, args) = op.parse_args()
Martin@648
   672
    if len(args) != 2: op.error("2 arguments needed")
Martin@648
   673
Martin@648
   674
    opts.text_color = "black"
Martin@648
   675
    opts.background_color = "white"
Martin@648
   676
    opts.label_space = 5     # minimum number of pixels between axis labels
Martin@648
   677
Martin@649
   678
    try: lastDotplot(opts, args)
Martin@649
   679
    except KeyboardInterrupt: pass  # avoid silly error message
Martin@649
   680
    except Exception, e:
Martin@649
   681
        prog = os.path.basename(sys.argv[0])
Martin@649
   682
        sys.exit(prog + ": error: " + str(e))