#! /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 gzip
import fnmatch, 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 == "-":
        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 croppedBlocks(blocks, range1, range2):
    cropBeg1, cropEnd1 = range1
    cropBeg2, cropEnd2 = range2
    if blocks[0][0] < 0: cropBeg1, cropEnd1 = -cropEnd1, -cropBeg1
    if blocks[0][1] < 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 seqRangeFromText(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 rangeFromSeqName(seqRanges, name, seqLen):
    if not seqRanges: return 0, seqLen
    base = name.split(".")[-1]  # allow for names like hg19.chr7
    for pat, beg, end in seqRanges:
        if fnmatch.fnmatchcase(name, pat) or fnmatch.fnmatchcase(base, pat):
            return max(beg, 0), min(end, seqLen)
    return None

def updateSeqs(isTrim, seqNames, seqLimits, seqName, seqRange, blocks, index):
    if seqName not in seqLimits:
        seqNames.append(seqName)
    if isTrim:
        beg = blocks[0][index]
        end = blocks[-1][index] + blocks[-1][2]
        if beg < 0: beg, end = -end, -beg
        if seqName in seqLimits:
            b, e = seqLimits[seqName]
            seqLimits[seqName] = min(b, beg), max(e, end)
        else:
            seqLimits[seqName] = beg, end
    else:
        seqLimits[seqName] = seqRange

def readAlignments(fileName, opts):
    '''Get alignments and sequence limits, from MAF or tabular format.'''
    seqRanges1 = map(seqRangeFromText, opts.seq1)
    seqRanges2 = map(seqRangeFromText, opts.seq2)

    alignments = []
    seqNames1 = []
    seqNames2 = []
    seqLimits1 = {}
    seqLimits2 = {}
    lines = myOpen(fileName)
    for seqName1, seqLen1, seqName2, seqLen2, blocks in alignmentInput(lines):
        range1 = rangeFromSeqName(seqRanges1, seqName1, seqLen1)
        if not range1: continue
        range2 = rangeFromSeqName(seqRanges2, seqName2, seqLen2)
        if not range2: continue
        b = list(croppedBlocks(list(blocks), range1, range2))
        if not b: continue
        aln = seqName1, seqName2, b
        alignments.append(aln)
        updateSeqs(opts.trim1, seqNames1, seqLimits1, seqName1, range1, b, 0)
        updateSeqs(opts.trim2, seqNames2, seqLimits2, seqName2, range2, b, 1)
    return alignments, seqNames1, seqNames2, seqLimits1, seqLimits2

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 textDimensions(imageDraw, font, textRot, text):
    x, y = imageDraw.textsize(text, font=font)
    return (y, x) if textRot else (x, y)

def get_text_sizes(my_strings, font, fontsize, image_mode, textRot):
    '''Get widths & heights, in pixels, of some strings.'''
    if fontsize == 0: return [(0, 0) for i in my_strings]
    image_size = 1, 1
    im = Image.new(image_mode, image_size)
    draw = ImageDraw.Draw(im)
    return [textDimensions(draw, font, textRot, i) for i in my_strings]

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 seqNameAndSizeText(seqName, seqSize):
    return seqName + ": " + sizeText(seqSize)

def getSeqInfo(sortOpt, seqNames, seqLimits,
               font, fontsize, image_mode, isShowSize, textRot):
    '''Return miscellaneous information about the sequences.'''
    if sortOpt == 1:
        seqNames.sort(key=natural_sort_key)
    seqSizes = [seqLimits[i][1] - seqLimits[i][0] for i in seqNames]
    for i in seqNames:
        r = seqLimits[i]
        out = i, str(r[0]), str(r[1])
        warn("\t".join(out))
    warn("")
    if sortOpt == 2:
        seqRecords = sorted(zip(seqSizes, seqNames), reverse=True)
        seqSizes = [i[0] for i in seqRecords]
        seqNames = [i[1] for i in seqRecords]
    if isShowSize:
        seqLabels = map(seqNameAndSizeText, seqNames, seqSizes)
    else:
        seqLabels = seqNames
    labelSizes = get_text_sizes(seqLabels, font, fontsize, image_mode, textRot)
    margin = max(i[1] for i in labelSizes)
    # xxx the margin may be too big, because some labels may get omitted
    return seqNames, seqSizes, seqLabels, labelSizes, 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(seq_sizes, pix_tween_seqs, pix_limit):
    '''Get the minimum bp-per-pixel that fits in the size limit.'''
    seq_num = len(seq_sizes)
    seq_pix_limit = pix_limit - pix_tween_seqs * (seq_num - 1)
    if seq_pix_limit < seq_num:
        raise Exception("can't fit the image: too many sequences?")
    negLimit = -seq_pix_limit
    negBpPerPix = sum(seq_sizes) // negLimit
    while True:
        if sum(i // negBpPerPix for i in seq_sizes) >= negLimit:
            return -negBpPerPix
        negBpPerPix -= 1

def get_seq_starts(seq_pix, pix_tween_seqs, margin):
    '''Get the start pixel for each sequence.'''
    seq_starts = []
    pix_tot = margin - pix_tween_seqs
    for i in seq_pix:
        pix_tot += pix_tween_seqs
        seq_starts.append(pix_tot)
        pix_tot += i
    return seq_starts

def get_pix_info(seq_sizes, bp_per_pix, pix_tween_seqs, margin):
    '''Return pixel information about the sequences.'''
    seq_pix = [div_ceil(i, bp_per_pix) for i in seq_sizes]
    seq_starts = get_seq_starts(seq_pix, pix_tween_seqs, margin)
    tot_pix = seq_starts[-1] + seq_pix[-1]
    return seq_pix, seq_starts, 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 alignmentPixels(width, height, alignments, bp_per_pix, origins1, origins2):
    hits = [0] * (width * height)  # the image data
    for seq1, seq2, blocks in alignments:
        ori1 = origins1[seq1]
        ori2 = origins2[seq2]
        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, seqLimits):
    if not fileName: return
    for line in myOpen(fileName):
        w = line.split()
        if not w: continue
        seqName = w[0]
        if seqName not in seqLimits: continue
        beg = int(w[1])
        end = int(w[2])
        layer = 900
        color = "#ffe4ff"
        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 = "#fff4f4"
                elif w[5] == "-":
                    color = "#f4f4ff"
        yield layer, color, seqName, beg, end

def commaSeparatedInts(text):
    return map(int, text.rstrip(",").split(","))

def readGenePred(opts, fileName, seqLimits):
    if not fileName: return
    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 seqLimits: 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, seqLimits):
    if not fileName: return
    for line in myOpen(fileName):
        fields = line.split()
        if len(fields) == 17:  # rmsk.txt
            seqName = fields[5]
            if seqName not in seqLimits: 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 seqLimits: 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, "#ffe4ff", seqName, beg, end
        elif strand == "+":
            yield 100, "#fff4f4", seqName, beg, end
        else:
            yield 100, "#f4f4ff", seqName, beg, end

def isExtraFirstGapField(fields):
    return fields[4].isdigit()

def readGaps(opts, fileName, seqLimits):
    '''Read locations of unsequenced gaps, from an agp or gap file.'''
    if not fileName: return
    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 seqLimits: 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, seqLimits, origins, margin, edge, isTop, bpPerPix):
    for layer, color, seqName, beg, end in beds:
        cropBeg, cropEnd = seqLimits[seqName]
        beg = max(beg, cropBeg)
        end = min(end, cropEnd)
        if beg >= end: continue
        ori = origins[seqName]
        if layer <= 1000:
            # include partly-covered pixels
            b = (ori + beg) // bpPerPix
            e = div_ceil(ori + end, bpPerPix)
        else:
            # exclude partly-covered pixels
            b = div_ceil(ori + beg, bpPerPix)
            e = (ori + end) // bpPerPix
            if e <= b: continue
        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 make_label(text, text_size, range_start, range_size):
    '''Return an axis label with endpoint & sort-order information.'''
    text_width, text_height = text_size
    label_start = range_start + (range_size - text_width) // 2
    label_end   = label_start + text_width
    sort_key    = text_width - range_size
    return sort_key, label_start, label_end, text, text_height

def get_nonoverlapping_labels(labels, label_space):
    '''Get a subset of non-overlapping axis labels, greedily.'''
    nonoverlapping_labels = []
    for i in labels:
        if True not in [i[1] < j[2] + label_space and j[1] < i[2] + label_space
                        for j in nonoverlapping_labels]:
            nonoverlapping_labels.append(i)
    return nonoverlapping_labels

def get_axis_image(seqNames, name_sizes, seq_starts, seq_pix, textRot,
                   textAln, font, image_mode, opts):
    '''Make an image of axis labels.'''
    min_pos = seq_starts[0]
    max_pos = seq_starts[-1] + seq_pix[-1]
    margin = max(i[1] for i in name_sizes)
    labels = map(make_label, seqNames, name_sizes, seq_starts, seq_pix)
    labels = [i for i in labels if i[1] >= min_pos and i[2] <= max_pos]
    labels.sort()
    minPixTweenLabels = 0 if textRot else opts.label_space
    labels = get_nonoverlapping_labels(labels, minPixTweenLabels)
    image_size = (margin, max_pos) if textRot else (max_pos, margin)
    im = Image.new(image_mode, image_size, opts.border_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 seqOrigins(seqNames, seq_starts, seqLimits, bp_per_pix):
    for i, j in zip(seqNames, seq_starts):
        yield i, bp_per_pix * j - seqLimits[i][0]

def lastDotplot(opts, args):
    if opts.fontfile:  font = ImageFont.truetype(opts.fontfile, opts.fontsize)
    else:              font = ImageFont.load_default()

    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])

    warn("reading alignments...")
    alignmentInfo = readAlignments(args[0], opts)
    alignments, seqNames1, seqNames2, seqLimits1, seqLimits2 = alignmentInfo
    warn("done")
    if not alignments: raise Exception("there are no alignments")

    textRot1 = "vertical".startswith(opts.rot1)
    i1 = getSeqInfo(opts.sort1, seqNames1, seqLimits1,
                    font, opts.fontsize, image_mode, opts.lengths1, textRot1)
    seqNames1, seqSizes1, seqLabels1, labelSizes1, tMargin = i1

    textRot2 = "horizontal".startswith(opts.rot2)
    i2 = getSeqInfo(opts.sort2, seqNames2, seqLimits2,
                    font, opts.fontsize, image_mode, opts.lengths2, textRot2)
    seqNames2, seqSizes2, seqLabels2, labelSizes2, lMargin = i2

    warn("choosing bp per pixel...")
    pix_limit1 = opts.width  - lMargin
    pix_limit2 = opts.height - tMargin
    bpPerPix1 = get_bp_per_pix(seqSizes1, opts.border_pixels, pix_limit1)
    bpPerPix2 = get_bp_per_pix(seqSizes2, opts.border_pixels, pix_limit2)
    bpPerPix = max(bpPerPix1, bpPerPix2)
    warn("bp per pixel = " + str(bpPerPix))

    seq_pix1, seq_starts1, width  = get_pix_info(seqSizes1, bpPerPix,
                                                 opts.border_pixels, lMargin)
    seq_pix2, seq_starts2, height = get_pix_info(seqSizes2, bpPerPix,
                                                 opts.border_pixels, tMargin)
    warn("width:  " + str(width))
    warn("height: " + str(height))

    origins1 = dict(seqOrigins(seqNames1, seq_starts1, seqLimits1, bpPerPix))
    origins2 = dict(seqOrigins(seqNames2, seq_starts2, seqLimits2, bpPerPix))

    warn("processing alignments...")
    hits = alignmentPixels(width, height, alignments, bpPerPix,
                           origins1, origins2)
    warn("done")

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

    seqLimits1 = expandedSeqDict(seqLimits1)
    seqLimits2 = expandedSeqDict(seqLimits2)
    origins1 = expandedSeqDict(origins1)
    origins2 = expandedSeqDict(origins2)

    beds1 = itertools.chain(readBed(opts.bed1, seqLimits1),
                            readRmsk(opts.rmsk1, seqLimits1),
                            readGenePred(opts, opts.genePred1, seqLimits1),
                            readGaps(opts, opts.gap1, seqLimits1))
    b1 = bedBoxes(beds1, seqLimits1, origins1, tMargin, height, True, bpPerPix)

    beds2 = itertools.chain(readBed(opts.bed2, seqLimits2),
                            readRmsk(opts.rmsk2, seqLimits2),
                            readGenePred(opts, opts.genePred2, seqLimits2),
                            readGaps(opts, opts.gap2, seqLimits2))
    b2 = bedBoxes(beds2, seqLimits2, origins2, lMargin, width, False, bpPerPix)

    boxes = sorted(itertools.chain(b1, b2))
    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 = get_axis_image(seqLabels1, labelSizes1, seq_starts1, seq_pix1,
                               textRot1, False, font, image_mode, opts)
        if textRot1:
            axis1 = axis1.transpose(Image.ROTATE_90)
        axis2 = get_axis_image(seqLabels2, labelSizes2, seq_starts2, seq_pix2,
                               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 seq_starts1[1:]:
        box = i - opts.border_pixels, tMargin, i, height
        im.paste(opts.border_color, box)

    for i in seq_starts2[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("--sort1", type="int", default=1, metavar="N",
                  help="genome1 sequence order: 0=input order, 1=name order, "
                  "2=length order (default=%default)")
    op.add_option("--sort2", type="int", default=1, metavar="N",
                  help="genome2 sequence order: 0=input order, 1=name order, "
                  "2=length order (default=%default)")
    op.add_option("--trim1", action="store_true",
                  help="trim unaligned sequence flanks from the 1st genome")
    op.add_option("--trim2", action="store_true",
                  help="trim unaligned sequence flanks from the 2nd genome")
    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="#dcdcdc",
                  help="color for pixels between sequences (default=%default)")
    # xxx --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=11,
                  help="TrueType or OpenType font size (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)")
    og.add_option("--lengths1", action="store_true",
                  help="show sequence lengths for the 1st (horizontal) genome")
    og.add_option("--lengths2", action="store_true",
                  help="show sequence lengths for the 2nd (vertical) genome")
    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="#dfd",
                  help="color for exons (default=%default)")
    og.add_option("--cds-color", metavar="COLOR", default="#bdb",
                  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="pink",
                  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, e:
        prog = os.path.basename(sys.argv[0])
        sys.exit(prog + ": error: " + str(e))
