def test_shorten(s, expected):
from jcvi.graphics.base import shorten
assert shorten(s) == expected, "Expect {}".format(expected)
def dotplot(args):
"""
%prog dotplot map.csv ref.fasta
Make dotplot between chromosomes and linkage maps.
The input map is csv formatted, for example:
ScaffoldID,ScaffoldPosition,LinkageGroup,GeneticPosition
scaffold_2707,11508,1,0
scaffold_2707,11525,1,1.2
"""
from jcvi.assembly.allmaps import CSVMapLine
from jcvi.formats.sizes import Sizes
from jcvi.utils.natsort import natsorted
from jcvi.graphics.base import shorten
from jcvi.graphics.dotplot import plt, savefig, markup, normalize_axes, \
downsample, plot_breaks_and_labels, thousands
p = OptionParser(dotplot.__doc__)
p.set_outfile(outfile=None)
opts, args, iopts = p.set_image_options(args,
figsize="8x8",
style="dark",
dpi=90,
cmap="copper")
if len(args) != 2:
sys.exit(not p.print_help())
csvfile, fastafile = args
sizes = natsorted(Sizes(fastafile).mapping.items())
seen = set()
raw_data = []
fig = plt.figure(1, (iopts.w, iopts.h))
root = fig.add_axes([0, 0, 1, 1]) # the whole canvas
ax = fig.add_axes([.1, .1, .8, .8]) # the dot plot
fp = must_open(csvfile)
for row in fp:
m = CSVMapLine(row)
seen.add(m.seqid)
raw_data.append(m)
# X-axis is the genome assembly
ctgs, ctg_sizes = zip(*sizes)
xsize = sum(ctg_sizes)
qb = list(np.cumsum(ctg_sizes))
qbreaks = list(zip(ctgs, [0] + qb, qb))
qstarts = dict(zip(ctgs, [0] + qb))
# Y-axis is the map
key = lambda x: x.lg
raw_data.sort(key=key)
ssizes = {}
for lg, d in groupby(raw_data, key=key):
ssizes[lg] = max([x.cm for x in d])
ssizes = natsorted(ssizes.items())
lgs, lg_sizes = zip(*ssizes)
ysize = sum(lg_sizes)
sb = list(np.cumsum(lg_sizes))
sbreaks = list(zip([("LG" + x) for x in lgs], [0] + sb, sb))
sstarts = dict(zip(lgs, [0] + sb))
# Re-code all the scatter dots
data = [(qstarts[x.seqid] + x.pos, sstarts[x.lg] + x.cm, 'g') \
for x in raw_data if (x.seqid in qstarts)]
npairs = downsample(data)
x, y, c = zip(*data)
ax.scatter(x, y, c=c, edgecolors="none", s=2, lw=0)
# Flip X-Y label
gy, gx = op.basename(csvfile).split(".")[:2]
gx, gy = shorten(gx, maxchar=30), shorten(gy, maxchar=30)
xlim, ylim = plot_breaks_and_labels(fig, root, ax, gx, gy, xsize, ysize,
qbreaks, sbreaks)
ax.set_xlim(xlim)
ax.set_ylim(ylim)
title = "Alignment: {} vs {}".format(gx, gy)
title += " ({} markers)".format(thousands(npairs))
root.set_title(markup(title), x=.5, y=.96, color="k")
logging.debug(title)
normalize_axes(root)
image_name = opts.outfile or \
(csvfile.rsplit(".", 1)[0] + "." + iopts.format)
savefig(image_name, dpi=iopts.dpi, iopts=iopts)
fig.clear()
def dotplot(args):
"""
%prog dotplot map.csv ref.fasta
Make dotplot between chromosomes and linkage maps.
The input map is csv formatted, for example:
ScaffoldID,ScaffoldPosition,LinkageGroup,GeneticPosition
scaffold_2707,11508,1,0
scaffold_2707,11525,1,1.2
"""
from jcvi.assembly.allmaps import CSVMapLine
from jcvi.formats.sizes import Sizes
from jcvi.utils.natsort import natsorted
from jcvi.graphics.base import shorten
from jcvi.graphics.dotplot import plt, savefig, markup, normalize_axes, \
downsample, plot_breaks_and_labels, thousands
p = OptionParser(dotplot.__doc__)
p.set_outfile(outfile=None)
opts, args, iopts = p.set_image_options(args, figsize="8x8",
style="dark", dpi=90, cmap="copper")
if len(args) != 2:
sys.exit(not p.print_help())
csvfile, fastafile = args
sizes = natsorted(Sizes(fastafile).mapping.items())
seen = set()
raw_data = []
fig = plt.figure(1, (iopts.w, iopts.h))
root = fig.add_axes([0, 0, 1, 1]) # the whole canvas
ax = fig.add_axes([.1, .1, .8, .8]) # the dot plot
fp = must_open(csvfile)
for row in fp:
m = CSVMapLine(row)
seen.add(m.seqid)
raw_data.append(m)
# X-axis is the genome assembly
ctgs, ctg_sizes = zip(*sizes)
xsize = sum(ctg_sizes)
qb = list(np.cumsum(ctg_sizes))
qbreaks = list(zip(ctgs, [0] + qb, qb))
qstarts = dict(zip(ctgs, [0] + qb))
# Y-axis is the map
key = lambda x: x.lg
raw_data.sort(key=key)
ssizes = {}
for lg, d in groupby(raw_data, key=key):
ssizes[lg] = max([x.cm for x in d])
ssizes = natsorted(ssizes.items())
lgs, lg_sizes = zip(*ssizes)
ysize = sum(lg_sizes)
sb = list(np.cumsum(lg_sizes))
sbreaks = list(zip([("LG" + x) for x in lgs], [0] + sb, sb))
sstarts = dict(zip(lgs, [0] + sb))
# Re-code all the scatter dots
data = [(qstarts[x.seqid] + x.pos, sstarts[x.lg] + x.cm, 'g') \
for x in raw_data if (x.seqid in qstarts)]
npairs = downsample(data)
x, y, c = zip(*data)
ax.scatter(x, y, c=c, edgecolors="none", s=2, lw=0)
# Flip X-Y label
gy, gx = op.basename(csvfile).split(".")[:2]
gx, gy = shorten(gx, maxchar=30), shorten(gy, maxchar=30)
xlim, ylim = plot_breaks_and_labels(fig, root, ax, gx, gy,
xsize, ysize, qbreaks, sbreaks)
ax.set_xlim(xlim)
ax.set_ylim(ylim)
title = "Alignment: {} vs {}".format(gx, gy)
title += " ({} markers)".format(thousands(npairs))
root.set_title(markup(title), x=.5, y=.96, color="k")
logging.debug(title)
normalize_axes(root)
image_name = opts.outfile or \
(csvfile.rsplit(".", 1)[0] + "." + iopts.format)
savefig(image_name, dpi=iopts.dpi, iopts=iopts)
fig.clear()
def plot(args):
"""
%prog plot input.bed seqid
Plot the matchings between the reconstructed pseudomolecules and the maps.
Two types of visualizations are available in one canvas:
1. Parallel axes, and matching markers are shown in connecting lines;
2. Scatter plot.
"""
from jcvi.graphics.base import plt, savefig, normalize_axes, \
set2, panel_labels, shorten
from jcvi.graphics.chromosome import Chromosome, GeneticMap, \
HorizontalChromosome
p = OptionParser(plot.__doc__)
add_allmaps_plot_options(p)
opts, args, iopts = p.set_image_options(args, figsize="10x6")
if len(args) != 2:
sys.exit(not p.print_help())
inputbed, seqid = args
pf = inputbed.rsplit(".", 1)[0]
bedfile = pf + ".lifted.bed"
agpfile = pf + ".agp"
weightsfile = opts.weightsfile
links = opts.links
function = get_function(opts.distance)
cc = Map(bedfile, function=function)
allseqids = cc.seqids
mapnames = cc.mapnames
weights = Weights(weightsfile, mapnames)
assert seqid in allseqids, "{0} not in {1}".format(seqid, allseqids)
s = Scaffold(seqid, cc)
mlgs = [k for k, v in s.mlg_counts.items() if v >= links]
while not mlgs:
links /= 2
logging.error("No markers to plot, --links reset to {0}".format(links))
mlgs = [k for k, v in s.mlg_counts.items() if v >= links]
mlgsizes = {}
for mlg in mlgs:
mm = cc.extract_mlg(mlg)
mlgsize = max(function(x) for x in mm)
mlgsizes[mlg] = mlgsize
fig = plt.figure(1, (iopts.w, iopts.h))
root = fig.add_axes([0, 0, 1, 1])
ax1 = fig.add_axes([0, 0, .5, 1])
ax2 = fig.add_axes([.5, 0, .5, 1])
# Find the layout first
ystart, ystop = .9, .1
L = Layout(mlgsizes)
coords = L.coords
tip = .02
marker_pos = {}
# Palette
colors = dict((mapname, set2[i]) for i, mapname in enumerate(mapnames))
colors = dict((mlg, colors[mlg.split("-")[0]]) for mlg in mlgs)
rhos = {}
# Parallel coordinates
for mlg, (x, y1, y2) in coords.items():
mm = cc.extract_mlg(mlg)
markers = [(m.accn, function(m)) for m in mm] # exhaustive marker list
xy = [(m.pos, function(m)) for m in mm if m.seqid == seqid]
mx, my = zip(*xy)
rho = spearmanr(mx, my)
rhos[mlg] = rho
flip = rho < 0
g = GeneticMap(ax1, x, y1, y2, markers, tip=tip, flip=flip)
extra = -3 * tip if x < .5 else 3 * tip
ha = "right" if x < .5 else "left"
mapname = mlg.split("-")[0]
tlg = shorten(mlg.replace("_", ".")) # Latex does not like underscore char
label = "{0} (w={1})".format(tlg, weights[mapname])
ax1.text(x + extra, (y1 + y2) / 2, label, color=colors[mlg],
ha=ha, va="center", rotation=90)
marker_pos.update(g.marker_pos)
agp = AGP(agpfile)
agp = [x for x in agp if x.object == seqid]
chrsize = max(x.object_end for x in agp)
# Pseudomolecules in the center
r = ystart - ystop
ratio = r / chrsize
f = lambda x: (ystart - ratio * x)
patchstart = [f(x.object_beg) for x in agp if not x.is_gap]
Chromosome(ax1, .5, ystart, ystop, width=2 * tip, patch=patchstart, lw=2)
label = "{0} ({1})".format(seqid, human_size(chrsize, precision=0))
ax1.text(.5, ystart + tip, label, ha="center")
scatter_data = defaultdict(list)
# Connecting lines
for b in s.markers:
marker_name = b.accn
if marker_name not in marker_pos:
continue
cx = .5
cy = f(b.pos)
mx = coords[b.mlg][0]
my = marker_pos[marker_name]
extra = -tip if mx < cx else tip
extra *= 1.25 # leave boundaries for aesthetic reasons
cx += extra
mx -= extra
ax1.plot((cx, mx), (cy, my), "-", color=colors[b.mlg])
scatter_data[b.mlg].append((b.pos, function(b)))
# Scatter plot, same data as parallel coordinates
xstart, xstop = sorted((ystart, ystop))
f = lambda x: (xstart + ratio * x)
pp = [x.object_beg for x in agp if not x.is_gap]
patchstart = [f(x) for x in pp]
HorizontalChromosome(ax2, xstart, xstop, ystop,
height=2 * tip, patch=patchstart, lw=2)
gap = .03
ratio = (r - gap * len(mlgs) - tip) / sum(mlgsizes.values())
tlgs = []
for mlg, mlgsize in sorted(mlgsizes.items()):
height = ratio * mlgsize
ystart -= height
xx = .5 + xstart / 2
width = r / 2
color = colors[mlg]
ax = fig.add_axes([xx, ystart, width, height])
ypos = ystart + height / 2
ystart -= gap
sd = scatter_data[mlg]
xx, yy = zip(*sd)
ax.vlines(pp, 0, 2 * mlgsize, colors="beige")
ax.plot(xx, yy, ".", color=color)
rho = rhos[mlg]
ax.text(.5, 1 - .4 * gap / height, r"$\rho$={0:.3f}".format(rho),
ha="center", va="top", transform=ax.transAxes, color="gray")
tlg = shorten(mlg.replace("_", "."))
tlgs.append((tlg, ypos, color))
ax.set_xlim(0, chrsize)
ax.set_ylim(0, mlgsize)
ax.set_xticks([])
while height / len(ax.get_yticks()) < .03 and len(ax.get_yticks()) >= 2:
ax.set_yticks(ax.get_yticks()[::2]) # Sparsify the ticks
yticklabels = [int(x) for x in ax.get_yticks()]
ax.set_yticklabels(yticklabels, family='Helvetica')
if rho < 0:
ax.invert_yaxis()
for i, (tlg, ypos, color) in enumerate(tlgs):
ha = "center"
if len(tlgs) > 4:
ha = "right" if i % 2 else "left"
root.text(.5, ypos, tlg, color=color, rotation=90,
ha=ha, va="center")
if opts.panels:
labels = ((.04, .96, 'A'), (.48, .96, 'B'))
panel_labels(root, labels)
normalize_axes((ax1, ax2, root))
image_name = seqid + "." + iopts.format
savefig(image_name, dpi=iopts.dpi, iopts=iopts)
plt.close(fig)