#! /usr/bin/env python

 # Read pair-wise alignments in MAF or LAST tabular format: write an

 # "Oxford grid", a.k.a. dotplot.

 # TODO: Currently, pixels with zero aligned nt-pairs are white, and

 # pixels with one or more aligned nt-pairs are black.  This can look

 # too crowded for large genome alignments.  I tried shading each pixel

 # according to the number of aligned nt-pairs within it, but the

 # result is too faint.  How can this be done better?

 import functools

 import gzip

 from fnmatch import fnmatchcase

 from operator import itemgetter

 import subprocess

 import itertools, optparse, os, re, sys

 # Try to make PIL/PILLOW work:

 try: from PIL import Image, ImageDraw, ImageFont, ImageColor

 except ImportError: import Image, ImageDraw, ImageFont, ImageColor

 def myOpen(fileName):  # faster than fileinput

     if fileName is None:

         return []

     if fileName == "-":

         return sys.stdin

     if fileName.endswith(".gz"):

         return gzip.open(fileName)

     return open(fileName)

 def warn(message):

     if opts.verbose:

         prog = os.path.basename(sys.argv[0])

         sys.stderr.write(prog + ": " + message + "\n")

 def groupByFirstItem(things):

     for k, v in itertools.groupby(things, itemgetter(0)):

         yield k, [i[1:] for i in v]

 def croppedBlocks(blocks, ranges1, ranges2):

     headBeg1, headBeg2, headSize = blocks[0]

     for r1 in ranges1:

         for r2 in ranges2:

             cropBeg1, cropEnd1 = r1

             if headBeg1 < 0:

                 cropBeg1, cropEnd1 = -cropEnd1, -cropBeg1

             cropBeg2, cropEnd2 = r2

             if headBeg2 < 0:

                 cropBeg2, cropEnd2 = -cropEnd2, -cropBeg2

             for beg1, beg2, size in blocks:

                 b1 = max(cropBeg1, beg1)

                 e1 = min(cropEnd1, beg1 + size)

                 if b1 >= e1: continue

                 offset = beg2 - beg1

                 b2 = max(cropBeg2, b1 + offset)

                 e2 = min(cropEnd2, e1 + offset)

                 if b2 >= e2: continue

                 yield b2 - offset, b2, e2 - b2

 def tabBlocks(beg1, beg2, blocks):

     '''Get the gapless blocks of an alignment, from LAST tabular format.'''

     for i in blocks.split(","):

         if ":" in i:

             x, y = i.split(":")

             beg1 += int(x)

             beg2 += int(y)

         else:

             size = int(i)

             yield beg1, beg2, size

             beg1 += size

             beg2 += size

 def mafBlocks(beg1, beg2, seq1, seq2):

     '''Get the gapless blocks of an alignment, from MAF format.'''

     size = 0

     for x, y in itertools.izip(seq1, seq2):

         if x == "-":

             if size:

                 yield beg1, beg2, size

                 beg1 += size

                 beg2 += size

                 size = 0

             beg2 += 1

         elif y == "-":

             if size:

                 yield beg1, beg2, size

                 beg1 += size

                 beg2 += size

                 size = 0

             beg1 += 1

         else:

             size += 1

     if size: yield beg1, beg2, size

 def alignmentInput(lines):

     '''Get alignments and sequence lengths, from MAF or tabular format.'''

     mafCount = 0

     for line in lines:

         w = line.split()

         if line[0].isdigit():  # tabular format

             chr1, beg1, seqlen1 = w[1], int(w[2]), int(w[5])

             if w[4] == "-": beg1 -= seqlen1

             chr2, beg2, seqlen2 = w[6], int(w[7]), int(w[10])

             if w[9] == "-": beg2 -= seqlen2

             blocks = tabBlocks(beg1, beg2, w[11])

             yield chr1, seqlen1, chr2, seqlen2, blocks

         elif line[0] == "s":  # MAF format

             if mafCount == 0:

                 chr1, beg1, seqlen1, seq1 = w[1], int(w[2]), int(w[5]), w[6]

                 if w[4] == "-": beg1 -= seqlen1

                 mafCount = 1

             else:

                 chr2, beg2, seqlen2, seq2 = w[1], int(w[2]), int(w[5]), w[6]

                 if w[4] == "-": beg2 -= seqlen2

                 blocks = mafBlocks(beg1, beg2, seq1, seq2)

                 yield chr1, seqlen1, chr2, seqlen2, blocks

                 mafCount = 0

 def seqRequestFromText(text):

     if ":" in text:

         pattern, interval = text.rsplit(":", 1)

         if "-" in interval:

             beg, end = interval.rsplit("-", 1)

             return pattern, int(beg), int(end)  # beg may be negative

     return text, 0, sys.maxsize

 def rangesFromSeqName(seqRequests, name, seqLen):

     if seqRequests:

         base = name.split(".")[-1]  # allow for names like hg19.chr7

         for pat, beg, end in seqRequests:

             if fnmatchcase(name, pat) or fnmatchcase(base, pat):

                 yield max(beg, 0), min(end, seqLen)

     else:

         yield 0, seqLen

 def updateSeqs(coverDict, seqRanges, seqName, ranges, coveredRange):

     beg, end = coveredRange

     if beg < 0:

         coveredRange = -end, -beg

     if seqName in coverDict:

         coverDict[seqName].append(coveredRange)

     else:

         coverDict[seqName] = [coveredRange]

         for beg, end in ranges:

             r = seqName, beg, end

             seqRanges.append(r)

 def readAlignments(fileName, opts):

     '''Get alignments and sequence limits, from MAF or tabular format.'''

     seqRequests1 = map(seqRequestFromText, opts.seq1)

     seqRequests2 = map(seqRequestFromText, opts.seq2)

     alignments = []

     seqRanges1 = []

     seqRanges2 = []

     coverDict1 = {}

     coverDict2 = {}

     lines = myOpen(fileName)

     for seqName1, seqLen1, seqName2, seqLen2, blocks in alignmentInput(lines):

         ranges1 = sorted(rangesFromSeqName(seqRequests1, seqName1, seqLen1))

         if not ranges1: continue

         ranges2 = sorted(rangesFromSeqName(seqRequests2, seqName2, seqLen2))

         if not ranges2: continue

         b = list(croppedBlocks(list(blocks), ranges1, ranges2))

         if not b: continue

         aln = seqName1, seqName2, b

         alignments.append(aln)

         coveredRange1 = b[0][0], b[-1][0] + b[-1][2]

         updateSeqs(coverDict1, seqRanges1, seqName1, ranges1, coveredRange1)

         coveredRange2 = b[0][1], b[-1][1] + b[-1][2]

         updateSeqs(coverDict2, seqRanges2, seqName2, ranges2, coveredRange2)

     return alignments, seqRanges1, coverDict1, seqRanges2, coverDict2

 def nameAndRangesFromDict(cropDict, seqName):

     if seqName in cropDict:

         return seqName, cropDict[seqName]

     n = seqName.split(".")[-1]

     if n in cropDict:

         return n, cropDict[n]

     return seqName, []

 def rangesForSecondaryAlignments(primaryRanges, seqLen):

     if primaryRanges:

         return primaryRanges

     return [(0, seqLen)]

 def readSecondaryAlignments(opts, cropRanges1, cropRanges2):

     cropDict1 = dict(groupByFirstItem(cropRanges1))

     cropDict2 = dict(groupByFirstItem(cropRanges2))

     alignments = []

     seqRanges1 = []

     seqRanges2 = []

     coverDict1 = {}

     coverDict2 = {}

     lines = myOpen(opts.alignments)

     for seqName1, seqLen1, seqName2, seqLen2, blocks in alignmentInput(lines):

         seqName1, ranges1 = nameAndRangesFromDict(cropDict1, seqName1)

         seqName2, ranges2 = nameAndRangesFromDict(cropDict2, seqName2)

         if not ranges1 and not ranges2:

             continue

         r1 = rangesForSecondaryAlignments(ranges1, seqLen1)

         r2 = rangesForSecondaryAlignments(ranges2, seqLen2)

         b = list(croppedBlocks(list(blocks), r1, r2))

         if not b: continue

         aln = seqName1, seqName2, b

         alignments.append(aln)

         if not ranges1:

             coveredRange1 = b[0][0], b[-1][0] + b[-1][2]

             updateSeqs(coverDict1, seqRanges1, seqName1, r1, coveredRange1)

         if not ranges2:

             coveredRange2 = b[0][1], b[-1][1] + b[-1][2]

             updateSeqs(coverDict2, seqRanges2, seqName2, r2, coveredRange2)

     return alignments, seqRanges1, coverDict1, seqRanges2, coverDict2

 def twoValuesFromOption(text, separator):

     if separator in text:

         return text.split(separator)

     return text, text

 def mergedRanges(ranges):

     oldBeg, maxEnd = ranges[0]

     for beg, end in ranges:

         if beg > maxEnd:

             yield oldBeg, maxEnd

             oldBeg = beg

             maxEnd = end

         elif end > maxEnd:

             maxEnd = end

     yield oldBeg, maxEnd

 def mergedRangesPerSeq(coverDict):

     for k, v in coverDict.iteritems():

         v.sort()

         yield k, list(mergedRanges(v))

 def coveredLength(mergedCoverDict):

     return sum(sum(e - b for b, e in v) for v in mergedCoverDict.itervalues())

 def trimmed(seqRanges, coverDict, minAlignedBases, maxGapFrac, endPad, midPad):

     maxEndGapFrac, maxMidGapFrac = twoValuesFromOption(maxGapFrac, ",")

     maxEndGap = max(float(maxEndGapFrac) * minAlignedBases, endPad * 1.0)

     maxMidGap = max(float(maxMidGapFrac) * minAlignedBases, midPad * 2.0)

     for seqName, rangeBeg, rangeEnd in seqRanges:

         seqBlocks = coverDict[seqName]

         blocks = [i for i in seqBlocks if i[0] < rangeEnd and i[1] > rangeBeg]

         if blocks[0][0] - rangeBeg > maxEndGap:

             rangeBeg = blocks[0][0] - endPad

         for j, y in enumerate(blocks):

             if j:

                 x = blocks[j - 1]

                 if y[0] - x[1] > maxMidGap:

                     yield seqName, rangeBeg, x[1] + midPad

                     rangeBeg = y[0] - midPad

         if rangeEnd - blocks[-1][1] > maxEndGap:

             rangeEnd = blocks[-1][1] + endPad

         yield seqName, rangeBeg, rangeEnd

 def natural_sort_key(my_string):

     '''Return a sort key for "natural" ordering, e.g. chr9 < chr10.'''

     parts = re.split(r'(\d+)', my_string)

     parts[1::2] = map(int, parts[1::2])

     return parts

 def nameKey(oneSeqRanges):

     return natural_sort_key(oneSeqRanges[0][0])

 def sizeKey(oneSeqRanges):

     return sum(b - e for n, b, e in oneSeqRanges), nameKey(oneSeqRanges)

 def alignmentKey(seqNamesToLists, oneSeqRanges):

     seqName = oneSeqRanges[0][0]

     alignmentsOfThisSequence = seqNamesToLists[seqName]

     numOfAlignedLetterPairs = sum(i[3] for i in alignmentsOfThisSequence)

     toMiddle = numOfAlignedLetterPairs // 2

     for i in alignmentsOfThisSequence:

         toMiddle -= i[3]

         if toMiddle < 0:

             return i[1:3]  # sequence-rank and "position" of this alignment

 def alignmentSortData1(alignments, seqNamesToRanks2):

     for seqName1, seqName2, blocks in alignments:

         seqRank2 = seqNamesToRanks2[seqName2]

         pos2 = abs(blocks[0][1] + blocks[-1][1] + blocks[-1][2])

         numOfAlignedLetterPairs = sum(i[2] for i in blocks)

         yield seqName1, seqRank2, pos2, numOfAlignedLetterPairs

 def alignmentSortData2(alignments, seqNamesToRanks1):

     for seqName1, seqName2, blocks in alignments:

         seqRank1 = seqNamesToRanks1[seqName1]

         pos1 = abs(blocks[0][0] + blocks[-1][0] + blocks[-1][2])

         numOfAlignedLetterPairs = sum(i[2] for i in blocks)

         yield seqName2, seqRank1, pos1, numOfAlignedLetterPairs

 def mySortedRanges(seqRanges, sortOpt,

                    alignmentSortDataFunc, alignments, otherRanges):

     rangesGroupedBySeqName = itertools.groupby(seqRanges, itemgetter(0))

     g = [list(ranges) for seqName, ranges in rangesGroupedBySeqName]

     if sortOpt == "1":

         g.sort(key=nameKey)

     if sortOpt == "2":

         g.sort(key=sizeKey)

     if sortOpt == "3":

         otherNameGroups = itertools.groupby(i[0] for i in otherRanges)

         ranksAndNames = enumerate(i[0] for i in otherNameGroups)

         otherNamesToRanks = dict((n, r) for r, n in ranksAndNames)

         alns = sorted(alignmentSortDataFunc(alignments, otherNamesToRanks))

         alnsGroupedBySeqName = itertools.groupby(alns, itemgetter(0))

         seqNamesToLists = dict((k, list(v)) for k, v in alnsGroupedBySeqName)

         g.sort(key=functools.partial(alignmentKey, seqNamesToLists))

     return [j for i in g for j in i]

 def allSortedRanges(opts, alignments, alignmentsB,

                     seqRanges1, seqRangesB1, seqRanges2, seqRangesB2):

     o1, oB1 = twoValuesFromOption(opts.sort1, ":")

     o2, oB2 = twoValuesFromOption(opts.sort2, ":")

     if o1 == "3" and o2 == "3":

         raise Exception("the sort options have circular dependency")

     if o1 != "3":

         s1 = mySortedRanges(seqRanges1, o1, None, None, None)

     if o2 != "3":

         s2 = mySortedRanges(seqRanges2, o2, None, None, None)

     if o1 == "3":

         s1 = mySortedRanges(seqRanges1, o1, alignmentSortData1, alignments, s2)

     if o2 == "3":

         s2 = mySortedRanges(seqRanges2, o2, alignmentSortData2, alignments, s1)

     t1 = mySortedRanges(seqRangesB1, oB1, alignmentSortData1, alignmentsB, s2)

     t2 = mySortedRanges(seqRangesB2, oB2, alignmentSortData2, alignmentsB, s1)

     return s1 + t1, s2 + t2

 def prettyNum(n):

     t = str(n)

     groups = []

     while t:

         groups.append(t[-3:])

         t = t[:-3]

     return ",".join(reversed(groups))

 def sizeText(size):

     suffixes = "bp", "kb", "Mb", "Gb"

     for i, x in enumerate(suffixes):

         j = 10 ** (i * 3)

         if size < j * 10:

             return "%.2g" % (1.0 * size / j) + x

         if size < j * 1000 or i == len(suffixes) - 1:

             return "%.0f" % (1.0 * size / j) + x

 def labelText(seqRange, labelOpt):

     seqName, beg, end = seqRange

     if labelOpt == 1:

         return seqName + ": " + sizeText(end - beg)

     if labelOpt == 2:

         return seqName + ":" + prettyNum(beg) + ": " + sizeText(end - beg)

     if labelOpt == 3:

         return seqName + ":" + prettyNum(beg) + "-" + prettyNum(end)

     return seqName

 def rangeLabels(seqRanges, labelOpt, font, fontsize, image_mode, textRot):

     if fontsize:

         image_size = 1, 1

         im = Image.new(image_mode, image_size)

         draw = ImageDraw.Draw(im)

     x = y = 0

     for r in seqRanges:

         text = labelText(r, labelOpt)

         if fontsize:

             x, y = draw.textsize(text, font=font)

             if textRot:

                 x, y = y, x

         yield text, x, y

 def dataFromRanges(sortedRanges, font, fontSize, imageMode, labelOpt, textRot):

     for i in sortedRanges:

         warn("\t".join(map(str, i)))

     warn("")

     rangeSizes = [e - b for n, b, e in sortedRanges]

     labs = list(rangeLabels(sortedRanges, labelOpt, font, fontSize,

                             imageMode, textRot))

     margin = max(i[2] for i in labs)

     # xxx the margin may be too big, because some labels may get omitted

     return rangeSizes, labs, margin

 def div_ceil(x, y):

     '''Return x / y rounded up.'''

     q, r = divmod(x, y)

     return q + (r != 0)

 def get_bp_per_pix(rangeSizes, pixTweenRanges, maxPixels):

     '''Get the minimum bp-per-pixel that fits in the size limit.'''

     warn("choosing bp per pixel...")

     numOfRanges = len(rangeSizes)

     maxPixelsInRanges = maxPixels - pixTweenRanges * (numOfRanges - 1)

     if maxPixelsInRanges < numOfRanges:

         raise Exception("can't fit the image: too many sequences?")

     negLimit = -maxPixelsInRanges

     negBpPerPix = sum(rangeSizes) // negLimit

     while True:

         if sum(i // negBpPerPix for i in rangeSizes) >= negLimit:

             return -negBpPerPix

         negBpPerPix -= 1

 def getRangePixBegs(rangePixLens, pixTweenRanges, margin):

     '''Get the start pixel for each range.'''

     rangePixBegs = []

     pix_tot = margin - pixTweenRanges

     for i in rangePixLens:

         pix_tot += pixTweenRanges

         rangePixBegs.append(pix_tot)

         pix_tot += i

     return rangePixBegs

 def pixelData(rangeSizes, bp_per_pix, pixTweenRanges, margin):

     '''Return pixel information about the ranges.'''

     rangePixLens = [div_ceil(i, bp_per_pix) for i in rangeSizes]

     rangePixBegs = getRangePixBegs(rangePixLens, pixTweenRanges, margin)

     tot_pix = rangePixBegs[-1] + rangePixLens[-1]

     return rangePixBegs, rangePixLens, tot_pix

 def drawLineForward(hits, width, bp_per_pix, beg1, beg2, size):

     while True:

         q1, r1 = divmod(beg1, bp_per_pix)

         q2, r2 = divmod(beg2, bp_per_pix)

         hits[q2 * width + q1] |= 1

         next_pix = min(bp_per_pix - r1, bp_per_pix - r2)

         if next_pix >= size: break

         beg1 += next_pix

         beg2 += next_pix

         size -= next_pix

 def drawLineReverse(hits, width, bp_per_pix, beg1, beg2, size):

     beg2 = -1 - beg2

     while True:

         q1, r1 = divmod(beg1, bp_per_pix)

         q2, r2 = divmod(beg2, bp_per_pix)

         hits[q2 * width + q1] |= 2

         next_pix = min(bp_per_pix - r1, r2 + 1)

         if next_pix >= size: break

         beg1 += next_pix

         beg2 -= next_pix

         size -= next_pix

 def findOrigin(ranges, beg, size):

     if beg < 0:

         beg = -(beg + size)

     for rangeBeg, rangeEnd, origin in ranges:

         if rangeEnd > beg:

             return origin

 def alignmentPixels(width, height, alignments, bp_per_pix,

                     rangeDict1, rangeDict2):

     hits = [0] * (width * height)  # the image data

     for seq1, seq2, blocks in alignments:

         beg1, beg2, size = blocks[0]

         ori1 = findOrigin(rangeDict1[seq1], beg1, size)

         ori2 = findOrigin(rangeDict2[seq2], beg2, size)

         for beg1, beg2, size in blocks:

             if beg1 < 0:

                 beg1 = -(beg1 + size)

                 beg2 = -(beg2 + size)

             if beg2 >= 0:

                 drawLineForward(hits, width, bp_per_pix,

                                 beg1 + ori1, beg2 + ori2, size)

             else:

                 drawLineReverse(hits, width, bp_per_pix,

                                 beg1 + ori1, beg2 - ori2, size)

     return hits

 def expandedSeqDict(seqDict):

     '''Allow lookup by short sequence names, e.g. chr7 as well as hg19.chr7.'''

     newDict = seqDict.copy()

     for name, x in seqDict.items():

         if "." in name:

             base = name.split(".")[-1]

             if base in newDict:  # an ambiguous case was found:

                 return seqDict   # so give up completely

             newDict[base] = x

     return newDict

 def readBed(fileName, rangeDict):

     for line in myOpen(fileName):

         w = line.split()

         if not w: continue

         seqName = w[0]

         if seqName not in rangeDict: continue

         beg = int(w[1])

         end = int(w[2])

         layer = 900

         color = "#fbf"

         if len(w) > 4:

             if w[4] != ".":

                 layer = float(w[4])

             if len(w) > 5:

                 if len(w) > 8 and w[8].count(",") == 2:

                     color = "rgb(" + w[8] + ")"

                 elif w[5] == "+":

                     color = "#ffe8e8"

                 elif w[5] == "-":

                     color = "#e8e8ff"

         yield layer, color, seqName, beg, end

 def commaSeparatedInts(text):

     return map(int, text.rstrip(",").split(","))

 def readGenePred(opts, fileName, rangeDict):

     for line in myOpen(fileName):

         fields = line.split()

         if not fields: continue

         if fields[2] not in "+-": fields = fields[1:]

         seqName = fields[1]

         if seqName not in rangeDict: continue

         #strand = fields[2]

         cdsBeg = int(fields[5])

         cdsEnd = int(fields[6])

         exonBegs = commaSeparatedInts(fields[8])

         exonEnds = commaSeparatedInts(fields[9])

         for beg, end in zip(exonBegs, exonEnds):

             yield 300, opts.exon_color, seqName, beg, end

             b = max(beg, cdsBeg)

             e = min(end, cdsEnd)

             if b < e: yield 400, opts.cds_color, seqName, b, e

 def readRmsk(fileName, rangeDict):

     for line in myOpen(fileName):

         fields = line.split()

         if len(fields) == 17:  # rmsk.txt

             seqName = fields[5]

             if seqName not in rangeDict: continue  # do this ASAP for speed

             beg = int(fields[6])

             end = int(fields[7])

             strand = fields[9]

             repeatClass = fields[11]

         elif len(fields) == 15:  # .out

             seqName = fields[4]

             if seqName not in rangeDict: continue

             beg = int(fields[5]) - 1

             end = int(fields[6])

             strand = fields[8]

             repeatClass = fields[10]

         else:

             continue

         if repeatClass in ("Low_complexity", "Simple_repeat"):

             yield 200, "#fbf", seqName, beg, end

         elif strand == "+":

             yield 100, "#ffe8e8", seqName, beg, end

         else:

             yield 100, "#e8e8ff", seqName, beg, end

 def isExtraFirstGapField(fields):

     return fields[4].isdigit()

 def readGaps(opts, fileName, rangeDict):

     '''Read locations of unsequenced gaps, from an agp or gap file.'''

     for line in myOpen(fileName):

         w = line.split()

         if not w or w[0][0] == "#": continue

         if isExtraFirstGapField(w): w = w[1:]

         if w[4] not in "NU": continue

         seqName = w[0]

         if seqName not in rangeDict: continue

         end = int(w[2])

         beg = end - int(w[5])  # zero-based coordinate

         if w[7] == "yes":

             yield 3000, opts.bridged_color, seqName, beg, end

         else:

             yield 2000, opts.unbridged_color, seqName, beg, end

 def bedBoxes(beds, rangeDict, margin, edge, isTop, bpPerPix):

     for layer, color, seqName, bedBeg, bedEnd in beds:

         for rangeBeg, rangeEnd, origin in rangeDict[seqName]:

             beg = max(bedBeg, rangeBeg)

             end = min(bedEnd, rangeEnd)

             if beg >= end: continue

             if layer <= 1000:

                 # include partly-covered pixels

                 b = (origin + beg) // bpPerPix

                 e = div_ceil(origin + end, bpPerPix)

             else:

                 # exclude partly-covered pixels

                 b = div_ceil(origin + beg, bpPerPix)

                 e = (origin + end) // bpPerPix

                 if e <= b: continue

                 if end == rangeEnd:  # include partly-covered end pixels

                     e = div_ceil(origin + end, bpPerPix)

             if isTop:

                 box = b, margin, e, edge

             else:

                 box = margin, b, edge, e

             yield layer, color, box

 def drawAnnotations(im, boxes):

     # xxx use partial transparency for different-color overlaps?

     for layer, color, box in boxes:

         im.paste(color, box)

 def placedLabels(labels, rangePixBegs, rangePixLens, beg, end):

     '''Return axis labels with endpoint & sort-order information.'''

     maxWidth = end - beg

     for i, j, k in zip(labels, rangePixBegs, rangePixLens):

         text, textWidth, textHeight = i

         if textWidth > maxWidth:

             continue

         labelBeg = j + (k - textWidth) // 2

         labelEnd = labelBeg + textWidth

         sortKey = textWidth - k

         if labelBeg < beg:

             sortKey += maxWidth * (beg - labelBeg)

             labelBeg = beg

             labelEnd = beg + textWidth

         if labelEnd > end:

             sortKey += maxWidth * (labelEnd - end)

             labelEnd = end

             labelBeg = end - textWidth

         yield sortKey, labelBeg, labelEnd, text, textHeight

 def nonoverlappingLabels(labels, minPixTweenLabels):

     '''Get a subset of non-overlapping axis labels, greedily.'''

     out = []

     for i in labels:

         beg = i[1] - minPixTweenLabels

         end = i[2] + minPixTweenLabels

         if all(j[2] <= beg or j[1] >= end for j in out):

             out.append(i)

     return out

 def axisImage(labels, rangePixBegs, rangePixLens, textRot,

               textAln, font, image_mode, opts):

     '''Make an image of axis labels.'''

     beg = rangePixBegs[0]

     end = rangePixBegs[-1] + rangePixLens[-1]

     margin = max(i[2] for i in labels)

     labels = sorted(placedLabels(labels, rangePixBegs, rangePixLens, beg, end))

     minPixTweenLabels = 0 if textRot else opts.label_space

     labels = nonoverlappingLabels(labels, minPixTweenLabels)

     image_size = (margin, end) if textRot else (end, margin)

     im = Image.new(image_mode, image_size, opts.margin_color)

     draw = ImageDraw.Draw(im)

     for i in labels:

         base = margin - i[4] if textAln else 0

         position = (base, i[1]) if textRot else (i[1], base)

         draw.text(position, i[3], font=font, fill=opts.text_color)

     return im

 def rangesWithOrigins(sortedRanges, rangePixBegs, bpPerPix):

     for i, j in zip(sortedRanges, rangePixBegs):

         seqName, rangeBeg, rangeEnd = i

         origin = bpPerPix * j - rangeBeg

         yield seqName, (rangeBeg, rangeEnd, origin)

 def rangesPerSeq(sortedRanges, rangePixBegs, bpPerPix):

     a = rangesWithOrigins(sortedRanges, rangePixBegs, bpPerPix)

     for k, v in itertools.groupby(a, itemgetter(0)):

         yield k, [i[1] for i in v]

 def getFont(opts):

     if opts.fontfile:

         return ImageFont.truetype(opts.fontfile, opts.fontsize)

     fileNames = []

     try:

         x = ["fc-match", "-f%{file}", "arial"]

         p = subprocess.Popen(x, stdout=subprocess.PIPE, stderr=subprocess.PIPE)

         out, err = p.communicate()

         fileNames.append(out)

     except OSError as e:

         warn("fc-match error: " + str(e))

     fileNames.append("/Library/Fonts/Arial.ttf")  # for Mac

     for i in fileNames:

         try:

             font = ImageFont.truetype(i, opts.fontsize)

             warn("font: " + i)

             return font

         except IOError as e:

             warn("font load error: " + str(e))

     return ImageFont.load_default()

 def lastDotplot(opts, args):

     font = getFont(opts)

     image_mode = 'RGB'

     forward_color = ImageColor.getcolor(opts.forwardcolor, image_mode)

     reverse_color = ImageColor.getcolor(opts.reversecolor, image_mode)

     zipped_colors = zip(forward_color, reverse_color)

     overlap_color = tuple([(i + j) // 2 for i, j in zipped_colors])

     maxGap1, maxGapB1 = twoValuesFromOption(opts.max_gap1, ":")

     maxGap2, maxGapB2 = twoValuesFromOption(opts.max_gap2, ":")

     warn("reading alignments...")

     alnData = readAlignments(args[0], opts)

     alignments, seqRanges1, coverDict1, seqRanges2, coverDict2 = alnData

     if not alignments: raise Exception("there are no alignments")

     warn("cutting...")

     coverDict1 = dict(mergedRangesPerSeq(coverDict1))

     coverDict2 = dict(mergedRangesPerSeq(coverDict2))

     minAlignedBases = min(coveredLength(coverDict1), coveredLength(coverDict2))

     pad = int(opts.pad * minAlignedBases)

     cutRanges1 = list(trimmed(seqRanges1, coverDict1, minAlignedBases,

                               maxGap1, pad, pad))

     cutRanges2 = list(trimmed(seqRanges2, coverDict2, minAlignedBases,

                               maxGap2, pad, pad))

     warn("reading secondary alignments...")

     alnDataB = readSecondaryAlignments(opts, cutRanges1, cutRanges2)

     alignmentsB, seqRangesB1, coverDictB1, seqRangesB2, coverDictB2 = alnDataB

     warn("cutting...")

     coverDictB1 = dict(mergedRangesPerSeq(coverDictB1))

     coverDictB2 = dict(mergedRangesPerSeq(coverDictB2))

     cutRangesB1 = trimmed(seqRangesB1, coverDictB1, minAlignedBases,

                           maxGapB1, 0, 0)

     cutRangesB2 = trimmed(seqRangesB2, coverDictB2, minAlignedBases,

                           maxGapB2, 0, 0)

     warn("sorting...")

     sortOut = allSortedRanges(opts, alignments, alignmentsB,

                               cutRanges1, cutRangesB1, cutRanges2, cutRangesB2)

     sortedRanges1, sortedRanges2 = sortOut

     textRot1 = "vertical".startswith(opts.rot1)

     i1 = dataFromRanges(sortedRanges1, font,

                         opts.fontsize, image_mode, opts.labels1, textRot1)

     rangeSizes1, labelData1, tMargin = i1

     textRot2 = "horizontal".startswith(opts.rot2)

     i2 = dataFromRanges(sortedRanges2, font,

                         opts.fontsize, image_mode, opts.labels2, textRot2)

     rangeSizes2, labelData2, lMargin = i2

     maxPixels1 = opts.width  - lMargin

     maxPixels2 = opts.height - tMargin

     bpPerPix1 = get_bp_per_pix(rangeSizes1, opts.border_pixels, maxPixels1)

     bpPerPix2 = get_bp_per_pix(rangeSizes2, opts.border_pixels, maxPixels2)

     bpPerPix = max(bpPerPix1, bpPerPix2)

     warn("bp per pixel = " + str(bpPerPix))

     p1 = pixelData(rangeSizes1, bpPerPix, opts.border_pixels, lMargin)

     rangePixBegs1, rangePixLens1, width = p1

     rangeDict1 = dict(rangesPerSeq(sortedRanges1, rangePixBegs1, bpPerPix))

     p2 = pixelData(rangeSizes2, bpPerPix, opts.border_pixels, tMargin)

     rangePixBegs2, rangePixLens2, height = p2

     rangeDict2 = dict(rangesPerSeq(sortedRanges2, rangePixBegs2, bpPerPix))

     warn("width:  " + str(width))

     warn("height: " + str(height))

     warn("processing alignments...")

     hits = alignmentPixels(width, height, alignments + alignmentsB, bpPerPix,

                            rangeDict1, rangeDict2)

     warn("reading annotations...")

     rangeDict1 = expandedSeqDict(rangeDict1)

     beds1 = itertools.chain(readBed(opts.bed1, rangeDict1),

                             readRmsk(opts.rmsk1, rangeDict1),

                             readGenePred(opts, opts.genePred1, rangeDict1),

                             readGaps(opts, opts.gap1, rangeDict1))

     b1 = bedBoxes(beds1, rangeDict1, tMargin, height, True, bpPerPix)

     rangeDict2 = expandedSeqDict(rangeDict2)

     beds2 = itertools.chain(readBed(opts.bed2, rangeDict2),

                             readRmsk(opts.rmsk2, rangeDict2),

                             readGenePred(opts, opts.genePred2, rangeDict2),

                             readGaps(opts, opts.gap2, rangeDict2))

     b2 = bedBoxes(beds2, rangeDict2, lMargin, width, False, bpPerPix)

     boxes = sorted(itertools.chain(b1, b2))

     warn("drawing...")

     image_size = width, height

     im = Image.new(image_mode, image_size, opts.background_color)

     drawAnnotations(im, boxes)

     for i in range(height):

         for j in range(width):

             store_value = hits[i * width + j]

             xy = j, i

             if   store_value == 1: im.putpixel(xy, forward_color)

             elif store_value == 2: im.putpixel(xy, reverse_color)

             elif store_value == 3: im.putpixel(xy, overlap_color)

     if opts.fontsize != 0:

         axis1 = axisImage(labelData1, rangePixBegs1, rangePixLens1,

                           textRot1, False, font, image_mode, opts)

         if textRot1:

             axis1 = axis1.transpose(Image.ROTATE_90)

         axis2 = axisImage(labelData2, rangePixBegs2, rangePixLens2,

                           textRot2, textRot2, font, image_mode, opts)

         if not textRot2:

             axis2 = axis2.transpose(Image.ROTATE_270)

         im.paste(axis1, (0, 0))

         im.paste(axis2, (0, 0))

     for i in rangePixBegs1[1:]:

         box = i - opts.border_pixels, tMargin, i, height

         im.paste(opts.border_color, box)

     for i in rangePixBegs2[1:]:

         box = lMargin, i - opts.border_pixels, width, i

         im.paste(opts.border_color, box)

     im.save(args[1])

 if __name__ == "__main__":

     usage = """%prog --help

    or: %prog [options] maf-or-tab-alignments dotplot.png

    or: %prog [options] maf-or-tab-alignments dotplot.gif

    or: ..."""

     description = "Draw a dotplot of pair-wise sequence alignments in MAF or tabular format."

     op = optparse.OptionParser(usage=usage, description=description)

     op.add_option("-v", "--verbose", action="count",

                   help="show progress messages & data about the plot")

     op.add_option("-1", "--seq1", metavar="PATTERN", action="append",

                   default=[],

                   help="which sequences to show from the 1st genome")

     op.add_option("-2", "--seq2", metavar="PATTERN", action="append",

                   default=[],

                   help="which sequences to show from the 2nd genome")

     # Replace "width" & "height" with a single "length" option?

     op.add_option("-x", "--width", type="int", default=1000,

                   help="maximum width in pixels (default: %default)")

     op.add_option("-y", "--height", type="int", default=1000,

                   help="maximum height in pixels (default: %default)")

     op.add_option("-c", "--forwardcolor", metavar="COLOR", default="red",

                   help="color for forward alignments (default: %default)")

     op.add_option("-r", "--reversecolor", metavar="COLOR", default="blue",

                   help="color for reverse alignments (default: %default)")

     op.add_option("--alignments", metavar="FILE", help="secondary alignments")

     op.add_option("--sort1", default="1", metavar="N",

                   help="genome1 sequence order: 0=input order, 1=name order, "

                   "2=length order, 3=alignment order (default=%default)")

     op.add_option("--sort2", default="1", metavar="N",

                   help="genome2 sequence order: 0=input order, 1=name order, "

                   "2=length order, 3=alignment order (default=%default)")

     op.add_option("--max-gap1", metavar="FRAC", default="0.5,2", help=

                   "maximum unaligned (end,mid) gap in genome1: "

                   "fraction of aligned length (default=%default)")

     op.add_option("--max-gap2", metavar="FRAC", default="0.5,2", help=

                   "maximum unaligned (end,mid) gap in genome2: "

                   "fraction of aligned length (default=%default)")

     op.add_option("--pad", metavar="FRAC", type="float", default=0.04, help=

                   "pad length when cutting unaligned gaps: "

                   "fraction of aligned length (default=%default)")

     op.add_option("--border-pixels", metavar="INT", type="int", default=1,

                   help="number of pixels between sequences (default=%default)")

     op.add_option("--border-color", metavar="COLOR", default="black",

                   help="color for pixels between sequences (default=%default)")

     # --break-color and/or --break-pixels for intra-sequence breaks?

     op.add_option("--margin-color", metavar="COLOR", default="#dcdcdc",

                   help="margin color")

     og = optparse.OptionGroup(op, "Text options")

     og.add_option("-f", "--fontfile", metavar="FILE",

                   help="TrueType or OpenType font file")

     og.add_option("-s", "--fontsize", metavar="SIZE", type="int", default=14,

                   help="TrueType or OpenType font size (default: %default)")

     og.add_option("--labels1", type="int", default=0, metavar="N", help=

                   "genome1 labels: 0=name, 1=name:length, "

                   "2=name:start:length, 3=name:start-end (default=%default)")

     og.add_option("--labels2", type="int", default=0, metavar="N", help=

                   "genome2 labels: 0=name, 1=name:length, "

                   "2=name:start:length, 3=name:start-end (default=%default)")

     og.add_option("--rot1", metavar="ROT", default="h",

                   help="text rotation for the 1st genome (default=%default)")

     og.add_option("--rot2", metavar="ROT", default="v",

                   help="text rotation for the 2nd genome (default=%default)")

     op.add_option_group(og)

     og = optparse.OptionGroup(op, "Annotation options")

     og.add_option("--bed1", metavar="FILE",

                   help="read genome1 annotations from BED file")

     og.add_option("--bed2", metavar="FILE",

                   help="read genome2 annotations from BED file")

     og.add_option("--rmsk1", metavar="FILE", help="read genome1 repeats from "

                   "RepeatMasker .out or rmsk.txt file")

     og.add_option("--rmsk2", metavar="FILE", help="read genome2 repeats from "

                   "RepeatMasker .out or rmsk.txt file")

     op.add_option_group(og)

     og = optparse.OptionGroup(op, "Gene options")

     og.add_option("--genePred1", metavar="FILE",

                   help="read genome1 genes from genePred file")

     og.add_option("--genePred2", metavar="FILE",

                   help="read genome2 genes from genePred file")

     og.add_option("--exon-color", metavar="COLOR", default="PaleGreen",

                   help="color for exons (default=%default)")

     og.add_option("--cds-color", metavar="COLOR", default="LimeGreen",

                   help="color for protein-coding regions (default=%default)")

     op.add_option_group(og)

     og = optparse.OptionGroup(op, "Unsequenced gap options")

     og.add_option("--gap1", metavar="FILE",

                   help="read genome1 unsequenced gaps from agp or gap file")

     og.add_option("--gap2", metavar="FILE",

                   help="read genome2 unsequenced gaps from agp or gap file")

     og.add_option("--bridged-color", metavar="COLOR", default="yellow",

                   help="color for bridged gaps (default: %default)")

     og.add_option("--unbridged-color", metavar="COLOR", default="orange",

                   help="color for unbridged gaps (default: %default)")

     op.add_option_group(og)

     (opts, args) = op.parse_args()

     if len(args) != 2: op.error("2 arguments needed")

     opts.text_color = "black"

     opts.background_color = "white"

     opts.label_space = 5     # minimum number of pixels between axis labels

     try: lastDotplot(opts, args)

     except KeyboardInterrupt: pass  # avoid silly error message

     except Exception as e:

         prog = os.path.basename(sys.argv[0])

         sys.exit(prog + ": error: " + str(e))