scripts/last-dotplot
author Martin C. Frith
Wed Aug 21 14:32:10 2019 +0900 (11 months ago)
changeset 984 fb43f850eba2
parent 981 0e184d90d5da
child 1062 6d8d56a74b14
permissions -rwxr-xr-x
dotplot: change satellite-repeat color to purple
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@916
    12
import collections
Martin@914
    13
import functools
Martin@875
    14
import gzip
Martin@896
    15
from fnmatch import fnmatchcase
Martin@961
    16
import logging
Martin@906
    17
from operator import itemgetter
Martin@903
    18
import subprocess
Martin@896
    19
import itertools, optparse, os, re, sys
Martin@475
    20
Martin@475
    21
# Try to make PIL/PILLOW work:
Martin@938
    22
try:
Martin@938
    23
    from PIL import Image, ImageDraw, ImageFont, ImageColor
Martin@938
    24
except ImportError:
Martin@938
    25
    import Image, ImageDraw, ImageFont, ImageColor
Martin@938
    26
Martin@938
    27
try:
Martin@938
    28
    from future_builtins import zip
Martin@938
    29
except ImportError:
Martin@938
    30
    pass
Martin@1
    31
Martin@844
    32
def myOpen(fileName):  # faster than fileinput
Martin@904
    33
    if fileName is None:
Martin@904
    34
        return []
Martin@844
    35
    if fileName == "-":
Martin@844
    36
        return sys.stdin
Martin@875
    37
    if fileName.endswith(".gz"):
Martin@957
    38
        return gzip.open(fileName, "rt")  # xxx dubious for Python2
Martin@844
    39
    return open(fileName)
Martin@844
    40
Martin@911
    41
def groupByFirstItem(things):
Martin@911
    42
    for k, v in itertools.groupby(things, itemgetter(0)):
Martin@911
    43
        yield k, [i[1:] for i in v]
Martin@911
    44
Martin@908
    45
def croppedBlocks(blocks, ranges1, ranges2):
Martin@908
    46
    headBeg1, headBeg2, headSize = blocks[0]
Martin@908
    47
    for r1 in ranges1:
Martin@908
    48
        for r2 in ranges2:
Martin@908
    49
            cropBeg1, cropEnd1 = r1
Martin@908
    50
            if headBeg1 < 0:
Martin@908
    51
                cropBeg1, cropEnd1 = -cropEnd1, -cropBeg1
Martin@908
    52
            cropBeg2, cropEnd2 = r2
Martin@908
    53
            if headBeg2 < 0:
Martin@908
    54
                cropBeg2, cropEnd2 = -cropEnd2, -cropBeg2
Martin@908
    55
            for beg1, beg2, size in blocks:
Martin@908
    56
                b1 = max(cropBeg1, beg1)
Martin@908
    57
                e1 = min(cropEnd1, beg1 + size)
Martin@908
    58
                if b1 >= e1: continue
Martin@908
    59
                offset = beg2 - beg1
Martin@908
    60
                b2 = max(cropBeg2, b1 + offset)
Martin@908
    61
                e2 = min(cropEnd2, e1 + offset)
Martin@908
    62
                if b2 >= e2: continue
Martin@908
    63
                yield b2 - offset, b2, e2 - b2
Martin@840
    64
Martin@482
    65
def tabBlocks(beg1, beg2, blocks):
Martin@482
    66
    '''Get the gapless blocks of an alignment, from LAST tabular format.'''
Martin@482
    67
    for i in blocks.split(","):
Martin@482
    68
        if ":" in i:
Martin@482
    69
            x, y = i.split(":")
Martin@482
    70
            beg1 += int(x)
Martin@482
    71
            beg2 += int(y)
Martin@482
    72
        else:
Martin@482
    73
            size = int(i)
Martin@482
    74
            yield beg1, beg2, size
Martin@482
    75
            beg1 += size
Martin@482
    76
            beg2 += size
Martin@272
    77
Martin@482
    78
def mafBlocks(beg1, beg2, seq1, seq2):
Martin@482
    79
    '''Get the gapless blocks of an alignment, from MAF format.'''
Martin@482
    80
    size = 0
Martin@938
    81
    for x, y in zip(seq1, seq2):
Martin@482
    82
        if x == "-":
Martin@482
    83
            if size:
Martin@482
    84
                yield beg1, beg2, size
Martin@482
    85
                beg1 += size
Martin@482
    86
                beg2 += size
Martin@482
    87
                size = 0
Martin@482
    88
            beg2 += 1
Martin@482
    89
        elif y == "-":
Martin@482
    90
            if size:
Martin@482
    91
                yield beg1, beg2, size
Martin@482
    92
                beg1 += size
Martin@482
    93
                beg2 += size
Martin@482
    94
                size = 0
Martin@482
    95
            beg1 += 1
Martin@272
    96
        else:
Martin@482
    97
            size += 1
Martin@482
    98
    if size: yield beg1, beg2, size
Martin@272
    99
Martin@980
   100
def alignmentFromSegment(qrySeqName, qrySeqLen, segment):
Martin@980
   101
    refSeqLen = sys.maxsize  # XXX
Martin@980
   102
    refSeqName, refSeqBeg, qrySeqBeg, size = segment
Martin@980
   103
    block = refSeqBeg, qrySeqBeg, size
Martin@980
   104
    return refSeqName, refSeqLen, qrySeqName, qrySeqLen, [block]
Martin@980
   105
Martin@482
   106
def alignmentInput(lines):
Martin@482
   107
    '''Get alignments and sequence lengths, from MAF or tabular format.'''
Martin@482
   108
    mafCount = 0
Martin@980
   109
    qrySeqName = ""
Martin@980
   110
    segments = []
Martin@272
   111
    for line in lines:
Martin@272
   112
        w = line.split()
Martin@980
   113
        if line[0] == "#":
Martin@980
   114
            pass
Martin@980
   115
        elif len(w) == 1:
Martin@980
   116
            for i in segments:
Martin@980
   117
                yield alignmentFromSegment(qrySeqName, qrySeqLen, i)
Martin@980
   118
            qrySeqName = w[0]
Martin@980
   119
            qrySeqLen = 0
Martin@980
   120
            segments = []
Martin@980
   121
        elif len(w) == 2 and qrySeqName and w[1].isdigit():
Martin@980
   122
            qrySeqLen += int(w[1])
Martin@981
   123
        elif len(w) == 4 and qrySeqName and w[1].isdigit() and w[3].isdigit():
Martin@981
   124
            refSeqName, refSeqBeg, refSeqEnd = w[0], int(w[1]), int(w[3])
Martin@980
   125
            size = abs(refSeqEnd - refSeqBeg)
Martin@980
   126
            if refSeqBeg > refSeqEnd:
Martin@980
   127
                refSeqBeg = -refSeqBeg
Martin@980
   128
            segments.append((refSeqName, refSeqBeg, qrySeqLen, size))
Martin@980
   129
            qrySeqLen += size
Martin@980
   130
        elif line[0].isdigit():  # tabular format
Martin@482
   131
            chr1, beg1, seqlen1 = w[1], int(w[2]), int(w[5])
Martin@482
   132
            if w[4] == "-": beg1 -= seqlen1
Martin@482
   133
            chr2, beg2, seqlen2 = w[6], int(w[7]), int(w[10])
Martin@482
   134
            if w[9] == "-": beg2 -= seqlen2
Martin@847
   135
            blocks = tabBlocks(beg1, beg2, w[11])
Martin@482
   136
            yield chr1, seqlen1, chr2, seqlen2, blocks
Martin@272
   137
        elif line[0] == "s":  # MAF format
Martin@482
   138
            if mafCount == 0:
Martin@482
   139
                chr1, beg1, seqlen1, seq1 = w[1], int(w[2]), int(w[5]), w[6]
Martin@482
   140
                if w[4] == "-": beg1 -= seqlen1
Martin@482
   141
                mafCount = 1
Martin@482
   142
            else:
Martin@482
   143
                chr2, beg2, seqlen2, seq2 = w[1], int(w[2]), int(w[5]), w[6]
Martin@482
   144
                if w[4] == "-": beg2 -= seqlen2
Martin@847
   145
                blocks = mafBlocks(beg1, beg2, seq1, seq2)
Martin@482
   146
                yield chr1, seqlen1, chr2, seqlen2, blocks
Martin@482
   147
                mafCount = 0
Martin@980
   148
    for i in segments:
Martin@980
   149
        yield alignmentFromSegment(qrySeqName, qrySeqLen, i)
Martin@272
   150
Martin@897
   151
def seqRequestFromText(text):
Martin@840
   152
    if ":" in text:
Martin@840
   153
        pattern, interval = text.rsplit(":", 1)
Martin@840
   154
        if "-" in interval:
Martin@840
   155
            beg, end = interval.rsplit("-", 1)
Martin@840
   156
            return pattern, int(beg), int(end)  # beg may be negative
Martin@840
   157
    return text, 0, sys.maxsize
Martin@840
   158
Martin@909
   159
def rangesFromSeqName(seqRequests, name, seqLen):
Martin@908
   160
    if seqRequests:
Martin@908
   161
        base = name.split(".")[-1]  # allow for names like hg19.chr7
Martin@908
   162
        for pat, beg, end in seqRequests:
Martin@908
   163
            if fnmatchcase(name, pat) or fnmatchcase(base, pat):
Martin@909
   164
                yield max(beg, 0), min(end, seqLen)
Martin@908
   165
    else:
Martin@909
   166
        yield 0, seqLen
Martin@651
   167
Martin@909
   168
def updateSeqs(coverDict, seqRanges, seqName, ranges, coveredRange):
Martin@909
   169
    beg, end = coveredRange
Martin@909
   170
    if beg < 0:
Martin@909
   171
        coveredRange = -end, -beg
Martin@909
   172
    if seqName in coverDict:
Martin@909
   173
        coverDict[seqName].append(coveredRange)
Martin@839
   174
    else:
Martin@909
   175
        coverDict[seqName] = [coveredRange]
Martin@909
   176
        for beg, end in ranges:
Martin@909
   177
            r = seqName, beg, end
Martin@909
   178
            seqRanges.append(r)
Martin@839
   179
Martin@651
   180
def readAlignments(fileName, opts):
Martin@839
   181
    '''Get alignments and sequence limits, from MAF or tabular format.'''
Martin@937
   182
    seqRequests1 = [seqRequestFromText(i) for i in opts.seq1]
Martin@937
   183
    seqRequests2 = [seqRequestFromText(i) for i in opts.seq2]
Martin@840
   184
Martin@482
   185
    alignments = []
Martin@909
   186
    seqRanges1 = []
Martin@909
   187
    seqRanges2 = []
Martin@909
   188
    coverDict1 = {}
Martin@909
   189
    coverDict2 = {}
Martin@844
   190
    lines = myOpen(fileName)
Martin@838
   191
    for seqName1, seqLen1, seqName2, seqLen2, blocks in alignmentInput(lines):
Martin@909
   192
        ranges1 = sorted(rangesFromSeqName(seqRequests1, seqName1, seqLen1))
Martin@909
   193
        if not ranges1: continue
Martin@909
   194
        ranges2 = sorted(rangesFromSeqName(seqRequests2, seqName2, seqLen2))
Martin@909
   195
        if not ranges2: continue
Martin@909
   196
        b = list(croppedBlocks(list(blocks), ranges1, ranges2))
Martin@840
   197
        if not b: continue
Martin@840
   198
        aln = seqName1, seqName2, b
Martin@482
   199
        alignments.append(aln)
Martin@909
   200
        coveredRange1 = b[0][0], b[-1][0] + b[-1][2]
Martin@909
   201
        updateSeqs(coverDict1, seqRanges1, seqName1, ranges1, coveredRange1)
Martin@909
   202
        coveredRange2 = b[0][1], b[-1][1] + b[-1][2]
Martin@909
   203
        updateSeqs(coverDict2, seqRanges2, seqName2, ranges2, coveredRange2)
Martin@909
   204
    return alignments, seqRanges1, coverDict1, seqRanges2, coverDict2
Martin@1
   205
Martin@911
   206
def nameAndRangesFromDict(cropDict, seqName):
Martin@911
   207
    if seqName in cropDict:
Martin@911
   208
        return seqName, cropDict[seqName]
Martin@911
   209
    n = seqName.split(".")[-1]
Martin@911
   210
    if n in cropDict:
Martin@911
   211
        return n, cropDict[n]
Martin@911
   212
    return seqName, []
Martin@911
   213
Martin@911
   214
def rangesForSecondaryAlignments(primaryRanges, seqLen):
Martin@911
   215
    if primaryRanges:
Martin@911
   216
        return primaryRanges
Martin@911
   217
    return [(0, seqLen)]
Martin@911
   218
Martin@911
   219
def readSecondaryAlignments(opts, cropRanges1, cropRanges2):
Martin@911
   220
    cropDict1 = dict(groupByFirstItem(cropRanges1))
Martin@911
   221
    cropDict2 = dict(groupByFirstItem(cropRanges2))
Martin@911
   222
Martin@911
   223
    alignments = []
Martin@911
   224
    seqRanges1 = []
Martin@911
   225
    seqRanges2 = []
Martin@911
   226
    coverDict1 = {}
Martin@911
   227
    coverDict2 = {}
Martin@911
   228
    lines = myOpen(opts.alignments)
Martin@911
   229
    for seqName1, seqLen1, seqName2, seqLen2, blocks in alignmentInput(lines):
Martin@911
   230
        seqName1, ranges1 = nameAndRangesFromDict(cropDict1, seqName1)
Martin@911
   231
        seqName2, ranges2 = nameAndRangesFromDict(cropDict2, seqName2)
Martin@911
   232
        if not ranges1 and not ranges2:
Martin@911
   233
            continue
Martin@911
   234
        r1 = rangesForSecondaryAlignments(ranges1, seqLen1)
Martin@911
   235
        r2 = rangesForSecondaryAlignments(ranges2, seqLen2)
Martin@911
   236
        b = list(croppedBlocks(list(blocks), r1, r2))
Martin@911
   237
        if not b: continue
Martin@911
   238
        aln = seqName1, seqName2, b
Martin@911
   239
        alignments.append(aln)
Martin@911
   240
        if not ranges1:
Martin@911
   241
            coveredRange1 = b[0][0], b[-1][0] + b[-1][2]
Martin@911
   242
            updateSeqs(coverDict1, seqRanges1, seqName1, r1, coveredRange1)
Martin@911
   243
        if not ranges2:
Martin@911
   244
            coveredRange2 = b[0][1], b[-1][1] + b[-1][2]
Martin@911
   245
            updateSeqs(coverDict2, seqRanges2, seqName2, r2, coveredRange2)
Martin@911
   246
    return alignments, seqRanges1, coverDict1, seqRanges2, coverDict2
Martin@911
   247
Martin@910
   248
def twoValuesFromOption(text, separator):
Martin@910
   249
    if separator in text:
Martin@910
   250
        return text.split(separator)
Martin@910
   251
    return text, text
Martin@910
   252
Martin@910
   253
def mergedRanges(ranges):
Martin@910
   254
    oldBeg, maxEnd = ranges[0]
Martin@910
   255
    for beg, end in ranges:
Martin@910
   256
        if beg > maxEnd:
Martin@910
   257
            yield oldBeg, maxEnd
Martin@910
   258
            oldBeg = beg
Martin@910
   259
            maxEnd = end
Martin@910
   260
        elif end > maxEnd:
Martin@910
   261
            maxEnd = end
Martin@910
   262
    yield oldBeg, maxEnd
Martin@910
   263
Martin@910
   264
def mergedRangesPerSeq(coverDict):
Martin@941
   265
    for k, v in coverDict.items():
Martin@910
   266
        v.sort()
Martin@910
   267
        yield k, list(mergedRanges(v))
Martin@910
   268
Martin@910
   269
def coveredLength(mergedCoverDict):
Martin@941
   270
    return sum(sum(e - b for b, e in v) for v in mergedCoverDict.values())
Martin@910
   271
Martin@910
   272
def trimmed(seqRanges, coverDict, minAlignedBases, maxGapFrac, endPad, midPad):
Martin@910
   273
    maxEndGapFrac, maxMidGapFrac = twoValuesFromOption(maxGapFrac, ",")
Martin@910
   274
    maxEndGap = max(float(maxEndGapFrac) * minAlignedBases, endPad * 1.0)
Martin@910
   275
    maxMidGap = max(float(maxMidGapFrac) * minAlignedBases, midPad * 2.0)
Martin@910
   276
Martin@910
   277
    for seqName, rangeBeg, rangeEnd in seqRanges:
Martin@910
   278
        seqBlocks = coverDict[seqName]
Martin@910
   279
        blocks = [i for i in seqBlocks if i[0] < rangeEnd and i[1] > rangeBeg]
Martin@910
   280
        if blocks[0][0] - rangeBeg > maxEndGap:
Martin@910
   281
            rangeBeg = blocks[0][0] - endPad
Martin@910
   282
        for j, y in enumerate(blocks):
Martin@910
   283
            if j:
Martin@910
   284
                x = blocks[j - 1]
Martin@910
   285
                if y[0] - x[1] > maxMidGap:
Martin@910
   286
                    yield seqName, rangeBeg, x[1] + midPad
Martin@910
   287
                    rangeBeg = y[0] - midPad
Martin@910
   288
        if rangeEnd - blocks[-1][1] > maxEndGap:
Martin@910
   289
            rangeEnd = blocks[-1][1] + endPad
Martin@910
   290
        yield seqName, rangeBeg, rangeEnd
Martin@910
   291
Martin@916
   292
def rangesWithStrandInfo(seqRanges, strandOpt, alignments, seqIndex):
Martin@916
   293
    if strandOpt == "1":
Martin@916
   294
        forwardMinusReverse = collections.defaultdict(int)
Martin@916
   295
        for i in alignments:
Martin@916
   296
            blocks = i[2]
Martin@916
   297
            beg1, beg2, size = blocks[0]
Martin@916
   298
            numOfAlignedLetterPairs = sum(i[2] for i in blocks)
Martin@916
   299
            if (beg1 < 0) != (beg2 < 0):  # opposite-strand alignment
Martin@916
   300
                numOfAlignedLetterPairs *= -1
Martin@916
   301
            forwardMinusReverse[i[seqIndex]] += numOfAlignedLetterPairs
Martin@916
   302
    strandNum = 0
Martin@916
   303
    for seqName, beg, end in seqRanges:
Martin@916
   304
        if strandOpt == "1":
Martin@916
   305
            strandNum = 1 if forwardMinusReverse[seqName] >= 0 else 2
Martin@916
   306
        yield seqName, beg, end, strandNum
Martin@916
   307
Martin@1
   308
def natural_sort_key(my_string):
Martin@1
   309
    '''Return a sort key for "natural" ordering, e.g. chr9 < chr10.'''
Martin@1
   310
    parts = re.split(r'(\d+)', my_string)
Martin@1
   311
    parts[1::2] = map(int, parts[1::2])
Martin@1
   312
    return parts
Martin@1
   313
Martin@907
   314
def nameKey(oneSeqRanges):
Martin@907
   315
    return natural_sort_key(oneSeqRanges[0][0])
Martin@907
   316
Martin@907
   317
def sizeKey(oneSeqRanges):
Martin@925
   318
    return sum(b - e for n, b, e, s in oneSeqRanges), nameKey(oneSeqRanges)
Martin@907
   319
Martin@914
   320
def alignmentKey(seqNamesToLists, oneSeqRanges):
Martin@914
   321
    seqName = oneSeqRanges[0][0]
Martin@914
   322
    alignmentsOfThisSequence = seqNamesToLists[seqName]
Martin@914
   323
    numOfAlignedLetterPairs = sum(i[3] for i in alignmentsOfThisSequence)
Martin@914
   324
    toMiddle = numOfAlignedLetterPairs // 2
Martin@914
   325
    for i in alignmentsOfThisSequence:
Martin@914
   326
        toMiddle -= i[3]
Martin@914
   327
        if toMiddle < 0:
Martin@914
   328
            return i[1:3]  # sequence-rank and "position" of this alignment
Martin@914
   329
Martin@916
   330
def rankAndFlipPerSeq(seqRanges):
Martin@916
   331
    rangesGroupedBySeqName = itertools.groupby(seqRanges, itemgetter(0))
Martin@916
   332
    for rank, group in enumerate(rangesGroupedBySeqName):
Martin@916
   333
        seqName, ranges = group
Martin@916
   334
        strandNum = next(ranges)[3]
Martin@916
   335
        flip = 1 if strandNum < 2 else -1
Martin@916
   336
        yield seqName, (rank, flip)
Martin@916
   337
Martin@916
   338
def alignmentSortData(alignments, seqIndex, otherNamesToRanksAndFlips):
Martin@915
   339
    otherIndex = 1 - seqIndex
Martin@915
   340
    for i in alignments:
Martin@915
   341
        blocks = i[2]
Martin@916
   342
        otherRank, otherFlip = otherNamesToRanksAndFlips[i[otherIndex]]
Martin@916
   343
        otherPos = otherFlip * abs(blocks[0][otherIndex] +
Martin@916
   344
                                   blocks[-1][otherIndex] + blocks[-1][2])
Martin@914
   345
        numOfAlignedLetterPairs = sum(i[2] for i in blocks)
Martin@915
   346
        yield i[seqIndex], otherRank, otherPos, numOfAlignedLetterPairs
Martin@914
   347
Martin@915
   348
def mySortedRanges(seqRanges, sortOpt, seqIndex, alignments, otherRanges):
Martin@913
   349
    rangesGroupedBySeqName = itertools.groupby(seqRanges, itemgetter(0))
Martin@913
   350
    g = [list(ranges) for seqName, ranges in rangesGroupedBySeqName]
Martin@916
   351
    for i in g:
Martin@916
   352
        if i[0][3] > 1:
Martin@916
   353
            i.reverse()
Martin@907
   354
    if sortOpt == "1":
Martin@907
   355
        g.sort(key=nameKey)
Martin@907
   356
    if sortOpt == "2":
Martin@907
   357
        g.sort(key=sizeKey)
Martin@914
   358
    if sortOpt == "3":
Martin@916
   359
        otherNamesToRanksAndFlips = dict(rankAndFlipPerSeq(otherRanges))
Martin@915
   360
        alns = sorted(alignmentSortData(alignments, seqIndex,
Martin@916
   361
                                        otherNamesToRanksAndFlips))
Martin@914
   362
        alnsGroupedBySeqName = itertools.groupby(alns, itemgetter(0))
Martin@914
   363
        seqNamesToLists = dict((k, list(v)) for k, v in alnsGroupedBySeqName)
Martin@914
   364
        g.sort(key=functools.partial(alignmentKey, seqNamesToLists))
Martin@907
   365
    return [j for i in g for j in i]
Martin@907
   366
Martin@914
   367
def allSortedRanges(opts, alignments, alignmentsB,
Martin@914
   368
                    seqRanges1, seqRangesB1, seqRanges2, seqRangesB2):
Martin@916
   369
    o1, oB1 = twoValuesFromOption(opts.strands1, ":")
Martin@916
   370
    o2, oB2 = twoValuesFromOption(opts.strands2, ":")
Martin@916
   371
    if o1 == "1" and o2 == "1":
Martin@916
   372
        raise Exception("the strand options have circular dependency")
Martin@916
   373
    seqRanges1 = list(rangesWithStrandInfo(seqRanges1, o1, alignments, 0))
Martin@916
   374
    seqRanges2 = list(rangesWithStrandInfo(seqRanges2, o2, alignments, 1))
Martin@916
   375
    seqRangesB1 = list(rangesWithStrandInfo(seqRangesB1, oB1, alignmentsB, 0))
Martin@916
   376
    seqRangesB2 = list(rangesWithStrandInfo(seqRangesB2, oB2, alignmentsB, 1))
Martin@916
   377
Martin@914
   378
    o1, oB1 = twoValuesFromOption(opts.sort1, ":")
Martin@914
   379
    o2, oB2 = twoValuesFromOption(opts.sort2, ":")
Martin@914
   380
    if o1 == "3" and o2 == "3":
Martin@914
   381
        raise Exception("the sort options have circular dependency")
Martin@914
   382
    if o1 != "3":
Martin@914
   383
        s1 = mySortedRanges(seqRanges1, o1, None, None, None)
Martin@914
   384
    if o2 != "3":
Martin@914
   385
        s2 = mySortedRanges(seqRanges2, o2, None, None, None)
Martin@914
   386
    if o1 == "3":
Martin@915
   387
        s1 = mySortedRanges(seqRanges1, o1, 0, alignments, s2)
Martin@914
   388
    if o2 == "3":
Martin@915
   389
        s2 = mySortedRanges(seqRanges2, o2, 1, alignments, s1)
Martin@915
   390
    t1 = mySortedRanges(seqRangesB1, oB1, 0, alignmentsB, s2)
Martin@915
   391
    t2 = mySortedRanges(seqRangesB2, oB2, 1, alignmentsB, s1)
Martin@914
   392
    return s1 + t1, s2 + t2
Martin@911
   393
Martin@898
   394
def prettyNum(n):
Martin@898
   395
    t = str(n)
Martin@898
   396
    groups = []
Martin@898
   397
    while t:
Martin@898
   398
        groups.append(t[-3:])
Martin@898
   399
        t = t[:-3]
Martin@904
   400
    return ",".join(reversed(groups))
Martin@898
   401
Martin@846
   402
def sizeText(size):
Martin@846
   403
    suffixes = "bp", "kb", "Mb", "Gb"
Martin@846
   404
    for i, x in enumerate(suffixes):
Martin@846
   405
        j = 10 ** (i * 3)
Martin@846
   406
        if size < j * 10:
Martin@846
   407
            return "%.2g" % (1.0 * size / j) + x
Martin@846
   408
        if size < j * 1000 or i == len(suffixes) - 1:
Martin@846
   409
            return "%.0f" % (1.0 * size / j) + x
Martin@846
   410
Martin@898
   411
def labelText(seqRange, labelOpt):
Martin@916
   412
    seqName, beg, end, strandNum = seqRange
Martin@898
   413
    if labelOpt == 1:
Martin@898
   414
        return seqName + ": " + sizeText(end - beg)
Martin@898
   415
    if labelOpt == 2:
Martin@904
   416
        return seqName + ":" + prettyNum(beg) + ": " + sizeText(end - beg)
Martin@898
   417
    if labelOpt == 3:
Martin@904
   418
        return seqName + ":" + prettyNum(beg) + "-" + prettyNum(end)
Martin@898
   419
    return seqName
Martin@846
   420
Martin@899
   421
def rangeLabels(seqRanges, labelOpt, font, fontsize, image_mode, textRot):
Martin@899
   422
    if fontsize:
Martin@899
   423
        image_size = 1, 1
Martin@899
   424
        im = Image.new(image_mode, image_size)
Martin@899
   425
        draw = ImageDraw.Draw(im)
Martin@899
   426
    x = y = 0
Martin@899
   427
    for r in seqRanges:
Martin@899
   428
        text = labelText(r, labelOpt)
Martin@899
   429
        if fontsize:
Martin@899
   430
            x, y = draw.textsize(text, font=font)
Martin@899
   431
            if textRot:
Martin@899
   432
                x, y = y, x
Martin@916
   433
        yield text, x, y, r[3]
Martin@899
   434
Martin@914
   435
def dataFromRanges(sortedRanges, font, fontSize, imageMode, labelOpt, textRot):
Martin@916
   436
    for seqName, rangeBeg, rangeEnd, strandNum in sortedRanges:
Martin@916
   437
        out = [seqName, str(rangeBeg), str(rangeEnd)]
Martin@916
   438
        if strandNum > 0:
Martin@916
   439
            out.append(".+-"[strandNum])
Martin@961
   440
        logging.info("\t".join(out))
Martin@961
   441
    logging.info("")
Martin@916
   442
    rangeSizes = [e - b for n, b, e, s in sortedRanges]
Martin@913
   443
    labs = list(rangeLabels(sortedRanges, labelOpt, font, fontSize,
Martin@913
   444
                            imageMode, textRot))
Martin@907
   445
    margin = max(i[2] for i in labs)
Martin@878
   446
    # xxx the margin may be too big, because some labels may get omitted
Martin@914
   447
    return rangeSizes, labs, margin
Martin@1
   448
Martin@1
   449
def div_ceil(x, y):
Martin@1
   450
    '''Return x / y rounded up.'''
Martin@1
   451
    q, r = divmod(x, y)
Martin@1
   452
    return q + (r != 0)
Martin@1
   453
Martin@896
   454
def get_bp_per_pix(rangeSizes, pixTweenRanges, maxPixels):
Martin@1
   455
    '''Get the minimum bp-per-pixel that fits in the size limit.'''
Martin@961
   456
    logging.info("choosing bp per pixel...")
Martin@896
   457
    numOfRanges = len(rangeSizes)
Martin@896
   458
    maxPixelsInRanges = maxPixels - pixTweenRanges * (numOfRanges - 1)
Martin@896
   459
    if maxPixelsInRanges < numOfRanges:
Martin@649
   460
        raise Exception("can't fit the image: too many sequences?")
Martin@896
   461
    negLimit = -maxPixelsInRanges
Martin@896
   462
    negBpPerPix = sum(rangeSizes) // negLimit
Martin@863
   463
    while True:
Martin@896
   464
        if sum(i // negBpPerPix for i in rangeSizes) >= negLimit:
Martin@863
   465
            return -negBpPerPix
Martin@863
   466
        negBpPerPix -= 1
Martin@1
   467
Martin@900
   468
def getRangePixBegs(rangePixLens, pixTweenRanges, margin):
Martin@900
   469
    '''Get the start pixel for each range.'''
Martin@900
   470
    rangePixBegs = []
Martin@896
   471
    pix_tot = margin - pixTweenRanges
Martin@900
   472
    for i in rangePixLens:
Martin@896
   473
        pix_tot += pixTweenRanges
Martin@900
   474
        rangePixBegs.append(pix_tot)
Martin@1
   475
        pix_tot += i
Martin@900
   476
    return rangePixBegs
Martin@1
   477
Martin@896
   478
def pixelData(rangeSizes, bp_per_pix, pixTweenRanges, margin):
Martin@900
   479
    '''Return pixel information about the ranges.'''
Martin@900
   480
    rangePixLens = [div_ceil(i, bp_per_pix) for i in rangeSizes]
Martin@900
   481
    rangePixBegs = getRangePixBegs(rangePixLens, pixTweenRanges, margin)
Martin@900
   482
    tot_pix = rangePixBegs[-1] + rangePixLens[-1]
Martin@900
   483
    return rangePixBegs, rangePixLens, tot_pix
Martin@1
   484
Martin@835
   485
def drawLineForward(hits, width, bp_per_pix, beg1, beg2, size):
Martin@639
   486
    while True:
Martin@639
   487
        q1, r1 = divmod(beg1, bp_per_pix)
Martin@639
   488
        q2, r2 = divmod(beg2, bp_per_pix)
Martin@835
   489
        hits[q2 * width + q1] |= 1
Martin@639
   490
        next_pix = min(bp_per_pix - r1, bp_per_pix - r2)
Martin@639
   491
        if next_pix >= size: break
Martin@639
   492
        beg1 += next_pix
Martin@639
   493
        beg2 += next_pix
Martin@639
   494
        size -= next_pix
Martin@639
   495
Martin@835
   496
def drawLineReverse(hits, width, bp_per_pix, beg1, beg2, size):
Martin@639
   497
    while True:
Martin@639
   498
        q1, r1 = divmod(beg1, bp_per_pix)
Martin@639
   499
        q2, r2 = divmod(beg2, bp_per_pix)
Martin@835
   500
        hits[q2 * width + q1] |= 2
Martin@639
   501
        next_pix = min(bp_per_pix - r1, r2 + 1)
Martin@639
   502
        if next_pix >= size: break
Martin@639
   503
        beg1 += next_pix
Martin@639
   504
        beg2 -= next_pix
Martin@639
   505
        size -= next_pix
Martin@639
   506
Martin@916
   507
def strandAndOrigin(ranges, beg, size):
Martin@916
   508
    isReverseStrand = (beg < 0)
Martin@916
   509
    if isReverseStrand:
Martin@905
   510
        beg = -(beg + size)
Martin@916
   511
    for rangeBeg, rangeEnd, isReverseRange, origin in ranges:
Martin@926
   512
        if rangeEnd > beg:  # assumes the ranges are sorted
Martin@916
   513
            return (isReverseStrand != isReverseRange), origin
Martin@905
   514
Martin@905
   515
def alignmentPixels(width, height, alignments, bp_per_pix,
Martin@905
   516
                    rangeDict1, rangeDict2):
Martin@640
   517
    hits = [0] * (width * height)  # the image data
Martin@640
   518
    for seq1, seq2, blocks in alignments:
Martin@905
   519
        beg1, beg2, size = blocks[0]
Martin@916
   520
        isReverse1, ori1 = strandAndOrigin(rangeDict1[seq1], beg1, size)
Martin@916
   521
        isReverse2, ori2 = strandAndOrigin(rangeDict2[seq2], beg2, size)
Martin@640
   522
        for beg1, beg2, size in blocks:
Martin@916
   523
            if isReverse1:
Martin@640
   524
                beg1 = -(beg1 + size)
Martin@640
   525
                beg2 = -(beg2 + size)
Martin@916
   526
            if isReverse1 == isReverse2:
Martin@835
   527
                drawLineForward(hits, width, bp_per_pix,
Martin@915
   528
                                ori1 + beg1, ori2 + beg2, size)
Martin@640
   529
            else:
Martin@835
   530
                drawLineReverse(hits, width, bp_per_pix,
Martin@915
   531
                                ori1 + beg1, ori2 - beg2 - 1, size)
Martin@640
   532
    return hits
Martin@1
   533
Martin@980
   534
def orientedBlocks(alignments, seqIndex):
Martin@980
   535
    otherIndex = 1 - seqIndex
Martin@980
   536
    for a in alignments:
Martin@980
   537
        seq1, seq2, blocks = a
Martin@980
   538
        for b in blocks:
Martin@980
   539
            beg1, beg2, size = b
Martin@980
   540
            if b[seqIndex] < 0:
Martin@980
   541
                b = -(beg1 + size), -(beg2 + size), size
Martin@980
   542
            yield a[seqIndex], b[seqIndex], a[otherIndex], b[otherIndex], size
Martin@980
   543
Martin@980
   544
def drawJoins(im, alignments, bpPerPix, seqIndex, rangeDict1, rangeDict2):
Martin@980
   545
    blocks = orientedBlocks(alignments, seqIndex)
Martin@980
   546
    oldSeq1 = ""
Martin@980
   547
    for seq1, beg1, seq2, beg2, size in sorted(blocks):
Martin@980
   548
        isReverse1, ori1 = strandAndOrigin(rangeDict1[seq1], beg1, size)
Martin@980
   549
        isReverse2, ori2 = strandAndOrigin(rangeDict2[seq2], beg2, size)
Martin@980
   550
        end1 = beg1 + size - 1
Martin@980
   551
        end2 = beg2 + size - 1
Martin@980
   552
        if isReverse1:
Martin@980
   553
            beg1 = -(beg1 + 1)
Martin@980
   554
            end1 = -(end1 + 1)
Martin@980
   555
        if isReverse2:
Martin@980
   556
            beg2 = -(beg2 + 1)
Martin@980
   557
            end2 = -(end2 + 1)
Martin@980
   558
        newPix1 = (ori1 + beg1) // bpPerPix
Martin@980
   559
        newPix2 = (ori2 + beg2) // bpPerPix
Martin@980
   560
        if seq1 == oldSeq1:
Martin@980
   561
            lowerPix2 = min(oldPix2, newPix2)
Martin@980
   562
            upperPix2 = max(oldPix2, newPix2)
Martin@980
   563
            midPix1 = (oldPix1 + newPix1) // 2
Martin@980
   564
            if isReverse1:
Martin@980
   565
                midPix1 = (oldPix1 + newPix1 + 1) // 2
Martin@980
   566
                oldPix1, newPix1 = newPix1, oldPix1
Martin@980
   567
            if upperPix2 - lowerPix2 > 1 and oldPix1 <= newPix1 <= oldPix1 + 1:
Martin@980
   568
                if seqIndex == 0:
Martin@980
   569
                    box = midPix1, lowerPix2, midPix1 + 1, upperPix2 + 1
Martin@980
   570
                else:
Martin@980
   571
                    box = lowerPix2, midPix1, upperPix2 + 1, midPix1 + 1
Martin@980
   572
                im.paste("lightgray", box)
Martin@980
   573
        oldPix1 = (ori1 + end1) // bpPerPix
Martin@980
   574
        oldPix2 = (ori2 + end2) // bpPerPix
Martin@980
   575
        oldSeq1 = seq1
Martin@980
   576
Martin@650
   577
def expandedSeqDict(seqDict):
Martin@650
   578
    '''Allow lookup by short sequence names, e.g. chr7 as well as hg19.chr7.'''
Martin@861
   579
    newDict = seqDict.copy()
Martin@650
   580
    for name, x in seqDict.items():
Martin@861
   581
        if "." in name:
Martin@861
   582
            base = name.split(".")[-1]
Martin@861
   583
            if base in newDict:  # an ambiguous case was found:
Martin@861
   584
                return seqDict   # so give up completely
Martin@861
   585
            newDict[base] = x
Martin@650
   586
    return newDict
Martin@650
   587
Martin@905
   588
def readBed(fileName, rangeDict):
Martin@845
   589
    for line in myOpen(fileName):
Martin@845
   590
        w = line.split()
Martin@847
   591
        if not w: continue
Martin@845
   592
        seqName = w[0]
Martin@905
   593
        if seqName not in rangeDict: continue
Martin@845
   594
        beg = int(w[1])
Martin@845
   595
        end = int(w[2])
Martin@857
   596
        layer = 900
Martin@895
   597
        color = "#fbf"
Martin@858
   598
        if len(w) > 4:
Martin@858
   599
            if w[4] != ".":
Martin@858
   600
                layer = float(w[4])
Martin@858
   601
            if len(w) > 5:
Martin@858
   602
                if len(w) > 8 and w[8].count(",") == 2:
Martin@858
   603
                    color = "rgb(" + w[8] + ")"
Martin@916
   604
                else:
Martin@916
   605
                    strand = w[5]
Martin@916
   606
                    isRev = rangeDict[seqName][0][2]
Martin@916
   607
                    if strand == "+" and not isRev or strand == "-" and isRev:
Martin@916
   608
                        color = "#ffe8e8"
Martin@916
   609
                    if strand == "-" and not isRev or strand == "+" and isRev:
Martin@916
   610
                        color = "#e8e8ff"
Martin@859
   611
        yield layer, color, seqName, beg, end
Martin@845
   612
Martin@860
   613
def commaSeparatedInts(text):
Martin@860
   614
    return map(int, text.rstrip(",").split(","))
Martin@860
   615
Martin@905
   616
def readGenePred(opts, fileName, rangeDict):
Martin@860
   617
    for line in myOpen(fileName):
Martin@860
   618
        fields = line.split()
Martin@860
   619
        if not fields: continue
Martin@860
   620
        if fields[2] not in "+-": fields = fields[1:]
Martin@860
   621
        seqName = fields[1]
Martin@905
   622
        if seqName not in rangeDict: continue
Martin@860
   623
        #strand = fields[2]
Martin@860
   624
        cdsBeg = int(fields[5])
Martin@860
   625
        cdsEnd = int(fields[6])
Martin@860
   626
        exonBegs = commaSeparatedInts(fields[8])
Martin@860
   627
        exonEnds = commaSeparatedInts(fields[9])
Martin@860
   628
        for beg, end in zip(exonBegs, exonEnds):
Martin@860
   629
            yield 300, opts.exon_color, seqName, beg, end
Martin@860
   630
            b = max(beg, cdsBeg)
Martin@860
   631
            e = min(end, cdsEnd)
Martin@860
   632
            if b < e: yield 400, opts.cds_color, seqName, b, e
Martin@860
   633
Martin@905
   634
def readRmsk(fileName, rangeDict):
Martin@860
   635
    for line in myOpen(fileName):
Martin@860
   636
        fields = line.split()
Martin@860
   637
        if len(fields) == 17:  # rmsk.txt
Martin@860
   638
            seqName = fields[5]
Martin@905
   639
            if seqName not in rangeDict: continue  # do this ASAP for speed
Martin@860
   640
            beg = int(fields[6])
Martin@860
   641
            end = int(fields[7])
Martin@860
   642
            strand = fields[9]
Martin@860
   643
            repeatClass = fields[11]
Martin@860
   644
        elif len(fields) == 15:  # .out
Martin@860
   645
            seqName = fields[4]
Martin@905
   646
            if seqName not in rangeDict: continue
Martin@860
   647
            beg = int(fields[5]) - 1
Martin@860
   648
            end = int(fields[6])
Martin@860
   649
            strand = fields[8]
Martin@984
   650
            repeatClass = fields[10].split("/")[0]
Martin@860
   651
        else:
Martin@860
   652
            continue
Martin@984
   653
        if repeatClass in ("Low_complexity", "Simple_repeat", "Satellite"):
Martin@895
   654
            yield 200, "#fbf", seqName, beg, end
Martin@916
   655
        elif (strand == "+") != rangeDict[seqName][0][2]:
Martin@895
   656
            yield 100, "#ffe8e8", seqName, beg, end
Martin@860
   657
        else:
Martin@895
   658
            yield 100, "#e8e8ff", seqName, beg, end
Martin@860
   659
Martin@650
   660
def isExtraFirstGapField(fields):
Martin@650
   661
    return fields[4].isdigit()
Martin@650
   662
Martin@905
   663
def readGaps(opts, fileName, rangeDict):
Martin@650
   664
    '''Read locations of unsequenced gaps, from an agp or gap file.'''
Martin@844
   665
    for line in myOpen(fileName):
Martin@650
   666
        w = line.split()
Martin@650
   667
        if not w or w[0][0] == "#": continue
Martin@650
   668
        if isExtraFirstGapField(w): w = w[1:]
Martin@650
   669
        if w[4] not in "NU": continue
Martin@650
   670
        seqName = w[0]
Martin@905
   671
        if seqName not in rangeDict: continue
Martin@650
   672
        end = int(w[2])
Martin@650
   673
        beg = end - int(w[5])  # zero-based coordinate
Martin@857
   674
        if w[7] == "yes":
Martin@859
   675
            yield 3000, opts.bridged_color, seqName, beg, end
Martin@857
   676
        else:
Martin@859
   677
            yield 2000, opts.unbridged_color, seqName, beg, end
Martin@650
   678
Martin@905
   679
def bedBoxes(beds, rangeDict, margin, edge, isTop, bpPerPix):
Martin@897
   680
    for layer, color, seqName, bedBeg, bedEnd in beds:
Martin@916
   681
        for rangeBeg, rangeEnd, isReverseRange, origin in rangeDict[seqName]:
Martin@905
   682
            beg = max(bedBeg, rangeBeg)
Martin@905
   683
            end = min(bedEnd, rangeEnd)
Martin@905
   684
            if beg >= end: continue
Martin@916
   685
            if isReverseRange:
Martin@916
   686
                beg, end = -end, -beg
Martin@905
   687
            if layer <= 1000:
Martin@905
   688
                # include partly-covered pixels
Martin@905
   689
                b = (origin + beg) // bpPerPix
Martin@905
   690
                e = div_ceil(origin + end, bpPerPix)
Martin@905
   691
            else:
Martin@905
   692
                # exclude partly-covered pixels
Martin@905
   693
                b = div_ceil(origin + beg, bpPerPix)
Martin@905
   694
                e = (origin + end) // bpPerPix
Martin@905
   695
                if e <= b: continue
Martin@916
   696
                if bedEnd >= rangeEnd:  # include partly-covered end pixels
Martin@916
   697
                    if isReverseRange:
Martin@916
   698
                        b = (origin + beg) // bpPerPix
Martin@916
   699
                    else:
Martin@916
   700
                        e = div_ceil(origin + end, bpPerPix)
Martin@905
   701
            if isTop:
Martin@905
   702
                box = b, margin, e, edge
Martin@905
   703
            else:
Martin@905
   704
                box = margin, b, edge, e
Martin@905
   705
            yield layer, color, box
Martin@845
   706
Martin@857
   707
def drawAnnotations(im, boxes):
Martin@857
   708
    # xxx use partial transparency for different-color overlaps?
Martin@857
   709
    for layer, color, box in boxes:
Martin@650
   710
        im.paste(color, box)
Martin@650
   711
Martin@901
   712
def placedLabels(labels, rangePixBegs, rangePixLens, beg, end):
Martin@901
   713
    '''Return axis labels with endpoint & sort-order information.'''
Martin@901
   714
    maxWidth = end - beg
Martin@901
   715
    for i, j, k in zip(labels, rangePixBegs, rangePixLens):
Martin@916
   716
        text, textWidth, textHeight, strandNum = i
Martin@901
   717
        if textWidth > maxWidth:
Martin@901
   718
            continue
Martin@901
   719
        labelBeg = j + (k - textWidth) // 2
Martin@901
   720
        labelEnd = labelBeg + textWidth
Martin@901
   721
        sortKey = textWidth - k
Martin@901
   722
        if labelBeg < beg:
Martin@904
   723
            sortKey += maxWidth * (beg - labelBeg)
Martin@901
   724
            labelBeg = beg
Martin@901
   725
            labelEnd = beg + textWidth
Martin@901
   726
        if labelEnd > end:
Martin@904
   727
            sortKey += maxWidth * (labelEnd - end)
Martin@901
   728
            labelEnd = end
Martin@901
   729
            labelBeg = end - textWidth
Martin@916
   730
        yield sortKey, labelBeg, labelEnd, text, textHeight, strandNum
Martin@1
   731
Martin@897
   732
def nonoverlappingLabels(labels, minPixTweenLabels):
Martin@1
   733
    '''Get a subset of non-overlapping axis labels, greedily.'''
Martin@897
   734
    out = []
Martin@1
   735
    for i in labels:
Martin@897
   736
        beg = i[1] - minPixTweenLabels
Martin@897
   737
        end = i[2] + minPixTweenLabels
Martin@897
   738
        if all(j[2] <= beg or j[1] >= end for j in out):
Martin@897
   739
            out.append(i)
Martin@897
   740
    return out
Martin@1
   741
Martin@901
   742
def axisImage(labels, rangePixBegs, rangePixLens, textRot,
Martin@895
   743
              textAln, font, image_mode, opts):
Martin@1
   744
    '''Make an image of axis labels.'''
Martin@900
   745
    beg = rangePixBegs[0]
Martin@900
   746
    end = rangePixBegs[-1] + rangePixLens[-1]
Martin@901
   747
    margin = max(i[2] for i in labels)
Martin@901
   748
    labels = sorted(placedLabels(labels, rangePixBegs, rangePixLens, beg, end))
Martin@878
   749
    minPixTweenLabels = 0 if textRot else opts.label_space
Martin@897
   750
    labels = nonoverlappingLabels(labels, minPixTweenLabels)
Martin@897
   751
    image_size = (margin, end) if textRot else (end, margin)
Martin@895
   752
    im = Image.new(image_mode, image_size, opts.margin_color)
Martin@1
   753
    draw = ImageDraw.Draw(im)
Martin@916
   754
    for sortKey, labelBeg, labelEnd, text, textHeight, strandNum in labels:
Martin@915
   755
        base = margin - textHeight if textAln else 0
Martin@915
   756
        position = (base, labelBeg) if textRot else (labelBeg, base)
Martin@916
   757
        fill = ("black", opts.forwardcolor, opts.reversecolor)[strandNum]
Martin@916
   758
        draw.text(position, text, font=font, fill=fill)
Martin@1
   759
    return im
Martin@1
   760
Martin@916
   761
def rangesWithOrigins(sortedRanges, rangePixBegs, rangePixLens, bpPerPix):
Martin@916
   762
    for i, j, k in zip(sortedRanges, rangePixBegs, rangePixLens):
Martin@916
   763
        seqName, rangeBeg, rangeEnd, strandNum = i
Martin@916
   764
        isReverseRange = (strandNum > 1)
Martin@916
   765
        if isReverseRange:
Martin@916
   766
            origin = bpPerPix * (j + k) + rangeBeg
Martin@916
   767
        else:
Martin@916
   768
            origin = bpPerPix * j - rangeBeg
Martin@916
   769
        yield seqName, (rangeBeg, rangeEnd, isReverseRange, origin)
Martin@906
   770
Martin@916
   771
def rangesPerSeq(sortedRanges, rangePixBegs, rangePixLens, bpPerPix):
Martin@916
   772
    a = rangesWithOrigins(sortedRanges, rangePixBegs, rangePixLens, bpPerPix)
Martin@906
   773
    for k, v in itertools.groupby(a, itemgetter(0)):
Martin@926
   774
        yield k, sorted(i[1] for i in v)
Martin@836
   775
Martin@903
   776
def getFont(opts):
Martin@903
   777
    if opts.fontfile:
Martin@903
   778
        return ImageFont.truetype(opts.fontfile, opts.fontsize)
Martin@903
   779
    fileNames = []
Martin@903
   780
    try:
Martin@903
   781
        x = ["fc-match", "-f%{file}", "arial"]
Martin@942
   782
        p = subprocess.Popen(x, stdout=subprocess.PIPE, stderr=subprocess.PIPE,
Martin@942
   783
                             universal_newlines=True)
Martin@903
   784
        out, err = p.communicate()
Martin@903
   785
        fileNames.append(out)
Martin@903
   786
    except OSError as e:
Martin@961
   787
        logging.info("fc-match error: " + str(e))
Martin@903
   788
    fileNames.append("/Library/Fonts/Arial.ttf")  # for Mac
Martin@903
   789
    for i in fileNames:
Martin@903
   790
        try:
Martin@903
   791
            font = ImageFont.truetype(i, opts.fontsize)
Martin@961
   792
            logging.info("font: " + i)
Martin@903
   793
            return font
Martin@903
   794
        except IOError as e:
Martin@961
   795
            logging.info("font load error: " + str(e))
Martin@903
   796
    return ImageFont.load_default()
Martin@903
   797
Martin@961
   798
def sequenceSizesAndNames(seqRanges):
Martin@961
   799
    for seqName, ranges in itertools.groupby(seqRanges, itemgetter(0)):
Martin@961
   800
        size = sum(e - b for n, b, e in ranges)
Martin@961
   801
        yield size, seqName
Martin@961
   802
Martin@961
   803
def biggestSequences(seqRanges, maxNumOfSequences):
Martin@961
   804
    s = sorted(sequenceSizesAndNames(seqRanges), reverse=True)
Martin@961
   805
    if len(s) > maxNumOfSequences:
Martin@961
   806
        logging.warning("too many sequences - discarding the smallest ones")
Martin@961
   807
        s = s[:maxNumOfSequences]
Martin@961
   808
    return set(i[1] for i in s)
Martin@961
   809
Martin@961
   810
def remainingSequenceRanges(seqRanges, alignments, seqIndex):
Martin@961
   811
    remainingSequences = set(i[seqIndex] for i in alignments)
Martin@961
   812
    return [i for i in seqRanges if i[0] in remainingSequences]
Martin@961
   813
Martin@648
   814
def lastDotplot(opts, args):
Martin@961
   815
    logLevel = logging.INFO if opts.verbose else logging.WARNING
Martin@961
   816
    logging.basicConfig(format="%(filename)s: %(message)s", level=logLevel)
Martin@961
   817
Martin@903
   818
    font = getFont(opts)
Martin@643
   819
    image_mode = 'RGB'
Martin@643
   820
    forward_color = ImageColor.getcolor(opts.forwardcolor, image_mode)
Martin@643
   821
    reverse_color = ImageColor.getcolor(opts.reversecolor, image_mode)
Martin@643
   822
    zipped_colors = zip(forward_color, reverse_color)
Martin@643
   823
    overlap_color = tuple([(i + j) // 2 for i, j in zipped_colors])
Martin@641
   824
Martin@911
   825
    maxGap1, maxGapB1 = twoValuesFromOption(opts.max_gap1, ":")
Martin@911
   826
    maxGap2, maxGapB2 = twoValuesFromOption(opts.max_gap2, ":")
Martin@911
   827
Martin@961
   828
    logging.info("reading alignments...")
Martin@904
   829
    alnData = readAlignments(args[0], opts)
Martin@909
   830
    alignments, seqRanges1, coverDict1, seqRanges2, coverDict2 = alnData
Martin@649
   831
    if not alignments: raise Exception("there are no alignments")
Martin@961
   832
    logging.info("cutting...")
Martin@910
   833
    coverDict1 = dict(mergedRangesPerSeq(coverDict1))
Martin@910
   834
    coverDict2 = dict(mergedRangesPerSeq(coverDict2))
Martin@910
   835
    minAlignedBases = min(coveredLength(coverDict1), coveredLength(coverDict2))
Martin@910
   836
    pad = int(opts.pad * minAlignedBases)
Martin@910
   837
    cutRanges1 = list(trimmed(seqRanges1, coverDict1, minAlignedBases,
Martin@911
   838
                              maxGap1, pad, pad))
Martin@910
   839
    cutRanges2 = list(trimmed(seqRanges2, coverDict2, minAlignedBases,
Martin@911
   840
                              maxGap2, pad, pad))
Martin@911
   841
Martin@961
   842
    biggestSeqs1 = biggestSequences(cutRanges1, opts.maxseqs)
Martin@961
   843
    cutRanges1 = [i for i in cutRanges1 if i[0] in biggestSeqs1]
Martin@961
   844
    alignments = [i for i in alignments if i[0] in biggestSeqs1]
Martin@961
   845
    cutRanges2 = remainingSequenceRanges(cutRanges2, alignments, 1)
Martin@961
   846
Martin@961
   847
    biggestSeqs2 = biggestSequences(cutRanges2, opts.maxseqs)
Martin@961
   848
    cutRanges2 = [i for i in cutRanges2 if i[0] in biggestSeqs2]
Martin@961
   849
    alignments = [i for i in alignments if i[1] in biggestSeqs2]
Martin@961
   850
    cutRanges1 = remainingSequenceRanges(cutRanges1, alignments, 0)
Martin@961
   851
Martin@961
   852
    logging.info("reading secondary alignments...")
Martin@911
   853
    alnDataB = readSecondaryAlignments(opts, cutRanges1, cutRanges2)
Martin@911
   854
    alignmentsB, seqRangesB1, coverDictB1, seqRangesB2, coverDictB2 = alnDataB
Martin@961
   855
    logging.info("cutting...")
Martin@911
   856
    coverDictB1 = dict(mergedRangesPerSeq(coverDictB1))
Martin@911
   857
    coverDictB2 = dict(mergedRangesPerSeq(coverDictB2))
Martin@911
   858
    cutRangesB1 = trimmed(seqRangesB1, coverDictB1, minAlignedBases,
Martin@911
   859
                          maxGapB1, 0, 0)
Martin@911
   860
    cutRangesB2 = trimmed(seqRangesB2, coverDictB2, minAlignedBases,
Martin@911
   861
                          maxGapB2, 0, 0)
Martin@641
   862
Martin@961
   863
    logging.info("sorting...")
Martin@914
   864
    sortOut = allSortedRanges(opts, alignments, alignmentsB,
Martin@914
   865
                              cutRanges1, cutRangesB1, cutRanges2, cutRangesB2)
Martin@914
   866
    sortedRanges1, sortedRanges2 = sortOut
Martin@914
   867
Martin@878
   868
    textRot1 = "vertical".startswith(opts.rot1)
Martin@914
   869
    i1 = dataFromRanges(sortedRanges1, font,
Martin@908
   870
                        opts.fontsize, image_mode, opts.labels1, textRot1)
Martin@914
   871
    rangeSizes1, labelData1, tMargin = i1
Martin@846
   872
Martin@878
   873
    textRot2 = "horizontal".startswith(opts.rot2)
Martin@914
   874
    i2 = dataFromRanges(sortedRanges2, font,
Martin@908
   875
                        opts.fontsize, image_mode, opts.labels2, textRot2)
Martin@914
   876
    rangeSizes2, labelData2, lMargin = i2
Martin@641
   877
Martin@896
   878
    maxPixels1 = opts.width  - lMargin
Martin@896
   879
    maxPixels2 = opts.height - tMargin
Martin@897
   880
    bpPerPix1 = get_bp_per_pix(rangeSizes1, opts.border_pixels, maxPixels1)
Martin@897
   881
    bpPerPix2 = get_bp_per_pix(rangeSizes2, opts.border_pixels, maxPixels2)
Martin@855
   882
    bpPerPix = max(bpPerPix1, bpPerPix2)
Martin@961
   883
    logging.info("bp per pixel = " + str(bpPerPix))
Martin@641
   884
Martin@900
   885
    p1 = pixelData(rangeSizes1, bpPerPix, opts.border_pixels, lMargin)
Martin@900
   886
    rangePixBegs1, rangePixLens1, width = p1
Martin@916
   887
    rangeDict1 = dict(rangesPerSeq(sortedRanges1, rangePixBegs1,
Martin@916
   888
                                   rangePixLens1, bpPerPix))
Martin@900
   889
Martin@900
   890
    p2 = pixelData(rangeSizes2, bpPerPix, opts.border_pixels, tMargin)
Martin@900
   891
    rangePixBegs2, rangePixLens2, height = p2
Martin@916
   892
    rangeDict2 = dict(rangesPerSeq(sortedRanges2, rangePixBegs2,
Martin@916
   893
                                   rangePixLens2, bpPerPix))
Martin@900
   894
Martin@961
   895
    logging.info("width:  " + str(width))
Martin@961
   896
    logging.info("height: " + str(height))
Martin@839
   897
Martin@961
   898
    logging.info("processing alignments...")
Martin@980
   899
    allAlignments = alignments + alignmentsB
Martin@980
   900
    hits = alignmentPixels(width, height, allAlignments, bpPerPix,
Martin@905
   901
                           rangeDict1, rangeDict2)
Martin@641
   902
Martin@961
   903
    logging.info("reading annotations...")
Martin@134
   904
Martin@905
   905
    rangeDict1 = expandedSeqDict(rangeDict1)
Martin@905
   906
    beds1 = itertools.chain(readBed(opts.bed1, rangeDict1),
Martin@905
   907
                            readRmsk(opts.rmsk1, rangeDict1),
Martin@905
   908
                            readGenePred(opts, opts.genePred1, rangeDict1),
Martin@905
   909
                            readGaps(opts, opts.gap1, rangeDict1))
Martin@905
   910
    b1 = bedBoxes(beds1, rangeDict1, tMargin, height, True, bpPerPix)
Martin@845
   911
Martin@905
   912
    rangeDict2 = expandedSeqDict(rangeDict2)
Martin@905
   913
    beds2 = itertools.chain(readBed(opts.bed2, rangeDict2),
Martin@905
   914
                            readRmsk(opts.rmsk2, rangeDict2),
Martin@905
   915
                            readGenePred(opts, opts.genePred2, rangeDict2),
Martin@905
   916
                            readGaps(opts, opts.gap2, rangeDict2))
Martin@905
   917
    b2 = bedBoxes(beds2, rangeDict2, lMargin, width, False, bpPerPix)
Martin@857
   918
Martin@857
   919
    boxes = sorted(itertools.chain(b1, b2))
Martin@904
   920
Martin@961
   921
    logging.info("drawing...")
Martin@904
   922
Martin@904
   923
    image_size = width, height
Martin@904
   924
    im = Image.new(image_mode, image_size, opts.background_color)
Martin@904
   925
Martin@857
   926
    drawAnnotations(im, boxes)
Martin@650
   927
Martin@980
   928
    joinA, joinB = twoValuesFromOption(opts.join, ":")
Martin@980
   929
    if joinA in "13":
Martin@980
   930
        drawJoins(im, alignments, bpPerPix, 0, rangeDict1, rangeDict2)
Martin@980
   931
    if joinB in "13":
Martin@980
   932
        drawJoins(im, alignmentsB, bpPerPix, 0, rangeDict1, rangeDict2)
Martin@980
   933
    if joinA in "23":
Martin@980
   934
        drawJoins(im, alignments, bpPerPix, 1, rangeDict2, rangeDict1)
Martin@980
   935
    if joinB in "23":
Martin@980
   936
        drawJoins(im, alignmentsB, bpPerPix, 1, rangeDict2, rangeDict1)
Martin@980
   937
Martin@643
   938
    for i in range(height):
Martin@643
   939
        for j in range(width):
Martin@643
   940
            store_value = hits[i * width + j]
Martin@643
   941
            xy = j, i
Martin@643
   942
            if   store_value == 1: im.putpixel(xy, forward_color)
Martin@643
   943
            elif store_value == 2: im.putpixel(xy, reverse_color)
Martin@643
   944
            elif store_value == 3: im.putpixel(xy, overlap_color)
Martin@95
   945
Martin@643
   946
    if opts.fontsize != 0:
Martin@900
   947
        axis1 = axisImage(labelData1, rangePixBegs1, rangePixLens1,
Martin@895
   948
                          textRot1, False, font, image_mode, opts)
Martin@878
   949
        if textRot1:
Martin@878
   950
            axis1 = axis1.transpose(Image.ROTATE_90)
Martin@900
   951
        axis2 = axisImage(labelData2, rangePixBegs2, rangePixLens2,
Martin@895
   952
                          textRot2, textRot2, font, image_mode, opts)
Martin@878
   953
        if not textRot2:
Martin@878
   954
            axis2 = axis2.transpose(Image.ROTATE_270)
Martin@643
   955
        im.paste(axis1, (0, 0))
Martin@643
   956
        im.paste(axis2, (0, 0))
Martin@1
   957
Martin@900
   958
    for i in rangePixBegs1[1:]:
Martin@877
   959
        box = i - opts.border_pixels, tMargin, i, height
Martin@852
   960
        im.paste(opts.border_color, box)
Martin@1
   961
Martin@900
   962
    for i in rangePixBegs2[1:]:
Martin@877
   963
        box = lMargin, i - opts.border_pixels, width, i
Martin@852
   964
        im.paste(opts.border_color, box)
Martin@1
   965
Martin@643
   966
    im.save(args[1])
Martin@648
   967
Martin@648
   968
if __name__ == "__main__":
Martin@649
   969
    usage = """%prog --help
Martin@649
   970
   or: %prog [options] maf-or-tab-alignments dotplot.png
Martin@649
   971
   or: %prog [options] maf-or-tab-alignments dotplot.gif
Martin@649
   972
   or: ..."""
Martin@649
   973
    description = "Draw a dotplot of pair-wise sequence alignments in MAF or tabular format."
Martin@649
   974
    op = optparse.OptionParser(usage=usage, description=description)
Martin@866
   975
    op.add_option("-v", "--verbose", action="count",
Martin@866
   976
                  help="show progress messages & data about the plot")
Martin@961
   977
    # Replace "width" & "height" with a single "length" option?
Martin@961
   978
    op.add_option("-x", "--width", metavar="INT", type="int", default=1000,
Martin@961
   979
                  help="maximum width in pixels (default: %default)")
Martin@961
   980
    op.add_option("-y", "--height", metavar="INT", type="int", default=1000,
Martin@961
   981
                  help="maximum height in pixels (default: %default)")
Martin@961
   982
    op.add_option("-m", "--maxseqs", type="int", default=100, metavar="M",
Martin@961
   983
                  help="maximum number of horizontal or vertical sequences "
Martin@961
   984
                  "(default=%default)")
Martin@651
   985
    op.add_option("-1", "--seq1", metavar="PATTERN", action="append",
Martin@840
   986
                  default=[],
Martin@651
   987
                  help="which sequences to show from the 1st genome")
Martin@651
   988
    op.add_option("-2", "--seq2", metavar="PATTERN", action="append",
Martin@840
   989
                  default=[],
Martin@651
   990
                  help="which sequences to show from the 2nd genome")
Martin@649
   991
    op.add_option("-c", "--forwardcolor", metavar="COLOR", default="red",
Martin@649
   992
                  help="color for forward alignments (default: %default)")
Martin@649
   993
    op.add_option("-r", "--reversecolor", metavar="COLOR", default="blue",
Martin@649
   994
                  help="color for reverse alignments (default: %default)")
Martin@911
   995
    op.add_option("--alignments", metavar="FILE", help="secondary alignments")
Martin@907
   996
    op.add_option("--sort1", default="1", metavar="N",
Martin@851
   997
                  help="genome1 sequence order: 0=input order, 1=name order, "
Martin@914
   998
                  "2=length order, 3=alignment order (default=%default)")
Martin@907
   999
    op.add_option("--sort2", default="1", metavar="N",
Martin@851
  1000
                  help="genome2 sequence order: 0=input order, 1=name order, "
Martin@914
  1001
                  "2=length order, 3=alignment order (default=%default)")
Martin@916
  1002
    op.add_option("--strands1", default="0", metavar="N", help=
Martin@916
  1003
                  "genome1 sequence orientation: 0=forward orientation, "
Martin@916
  1004
                  "1=alignment orientation (default=%default)")
Martin@916
  1005
    op.add_option("--strands2", default="0", metavar="N", help=
Martin@916
  1006
                  "genome2 sequence orientation: 0=forward orientation, "
Martin@916
  1007
                  "1=alignment orientation (default=%default)")
Martin@910
  1008
    op.add_option("--max-gap1", metavar="FRAC", default="0.5,2", help=
Martin@910
  1009
                  "maximum unaligned (end,mid) gap in genome1: "
Martin@910
  1010
                  "fraction of aligned length (default=%default)")
Martin@910
  1011
    op.add_option("--max-gap2", metavar="FRAC", default="0.5,2", help=
Martin@910
  1012
                  "maximum unaligned (end,mid) gap in genome2: "
Martin@910
  1013
                  "fraction of aligned length (default=%default)")
Martin@910
  1014
    op.add_option("--pad", metavar="FRAC", type="float", default=0.04, help=
Martin@910
  1015
                  "pad length when cutting unaligned gaps: "
Martin@910
  1016
                  "fraction of aligned length (default=%default)")
Martin@980
  1017
    op.add_option("-j", "--join", default="0", metavar="N", help=
Martin@980
  1018
                  "join: 0=nothing, 1=alignments adjacent in genome1, "
Martin@980
  1019
                  "2=alignments adjacent in genome2 (default=%default)")
Martin@852
  1020
    op.add_option("--border-pixels", metavar="INT", type="int", default=1,
Martin@852
  1021
                  help="number of pixels between sequences (default=%default)")
Martin@895
  1022
    op.add_option("--border-color", metavar="COLOR", default="black",
Martin@852
  1023
                  help="color for pixels between sequences (default=%default)")
Martin@911
  1024
    # --break-color and/or --break-pixels for intra-sequence breaks?
Martin@895
  1025
    op.add_option("--margin-color", metavar="COLOR", default="#dcdcdc",
Martin@895
  1026
                  help="margin color")
Martin@846
  1027
Martin@850
  1028
    og = optparse.OptionGroup(op, "Text options")
Martin@850
  1029
    og.add_option("-f", "--fontfile", metavar="FILE",
Martin@850
  1030
                  help="TrueType or OpenType font file")
Martin@903
  1031
    og.add_option("-s", "--fontsize", metavar="SIZE", type="int", default=14,
Martin@850
  1032
                  help="TrueType or OpenType font size (default: %default)")
Martin@898
  1033
    og.add_option("--labels1", type="int", default=0, metavar="N", help=
Martin@898
  1034
                  "genome1 labels: 0=name, 1=name:length, "
Martin@898
  1035
                  "2=name:start:length, 3=name:start-end (default=%default)")
Martin@898
  1036
    og.add_option("--labels2", type="int", default=0, metavar="N", help=
Martin@898
  1037
                  "genome2 labels: 0=name, 1=name:length, "
Martin@898
  1038
                  "2=name:start:length, 3=name:start-end (default=%default)")
Martin@878
  1039
    og.add_option("--rot1", metavar="ROT", default="h",
Martin@878
  1040
                  help="text rotation for the 1st genome (default=%default)")
Martin@878
  1041
    og.add_option("--rot2", metavar="ROT", default="v",
Martin@878
  1042
                  help="text rotation for the 2nd genome (default=%default)")
Martin@850
  1043
    op.add_option_group(og)
Martin@850
  1044
Martin@860
  1045
    og = optparse.OptionGroup(op, "Annotation options")
Martin@860
  1046
    og.add_option("--bed1", metavar="FILE",
Martin@860
  1047
                  help="read genome1 annotations from BED file")
Martin@860
  1048
    og.add_option("--bed2", metavar="FILE",
Martin@860
  1049
                  help="read genome2 annotations from BED file")
Martin@860
  1050
    og.add_option("--rmsk1", metavar="FILE", help="read genome1 repeats from "
Martin@860
  1051
                  "RepeatMasker .out or rmsk.txt file")
Martin@860
  1052
    og.add_option("--rmsk2", metavar="FILE", help="read genome2 repeats from "
Martin@860
  1053
                  "RepeatMasker .out or rmsk.txt file")
Martin@860
  1054
    op.add_option_group(og)
Martin@860
  1055
Martin@860
  1056
    og = optparse.OptionGroup(op, "Gene options")
Martin@860
  1057
    og.add_option("--genePred1", metavar="FILE",
Martin@860
  1058
                  help="read genome1 genes from genePred file")
Martin@860
  1059
    og.add_option("--genePred2", metavar="FILE",
Martin@860
  1060
                  help="read genome2 genes from genePred file")
Martin@895
  1061
    og.add_option("--exon-color", metavar="COLOR", default="PaleGreen",
Martin@860
  1062
                  help="color for exons (default=%default)")
Martin@895
  1063
    og.add_option("--cds-color", metavar="COLOR", default="LimeGreen",
Martin@860
  1064
                  help="color for protein-coding regions (default=%default)")
Martin@860
  1065
    op.add_option_group(og)
Martin@860
  1066
Martin@650
  1067
    og = optparse.OptionGroup(op, "Unsequenced gap options")
Martin@650
  1068
    og.add_option("--gap1", metavar="FILE",
Martin@650
  1069
                  help="read genome1 unsequenced gaps from agp or gap file")
Martin@650
  1070
    og.add_option("--gap2", metavar="FILE",
Martin@650
  1071
                  help="read genome2 unsequenced gaps from agp or gap file")
Martin@650
  1072
    og.add_option("--bridged-color", metavar="COLOR", default="yellow",
Martin@650
  1073
                  help="color for bridged gaps (default: %default)")
Martin@895
  1074
    og.add_option("--unbridged-color", metavar="COLOR", default="orange",
Martin@650
  1075
                  help="color for unbridged gaps (default: %default)")
Martin@650
  1076
    op.add_option_group(og)
Martin@648
  1077
    (opts, args) = op.parse_args()
Martin@648
  1078
    if len(args) != 2: op.error("2 arguments needed")
Martin@648
  1079
Martin@648
  1080
    opts.background_color = "white"
Martin@648
  1081
    opts.label_space = 5     # minimum number of pixels between axis labels
Martin@648
  1082
Martin@649
  1083
    try: lastDotplot(opts, args)
Martin@649
  1084
    except KeyboardInterrupt: pass  # avoid silly error message
Martin@903
  1085
    except Exception as e:
Martin@649
  1086
        prog = os.path.basename(sys.argv[0])
Martin@649
  1087
        sys.exit(prog + ": error: " + str(e))