def model_makeplots(indir, outdir):
objs = lcommon.load_pickledir(indir)
if not objs:
return
rcommon.union_models(objs)
for attr in objs[0].get_attrs_depth1():
outbase = os.path.join(outdir, attr)
k2ydata = model_get_ydata(objs, attr)
gene_attrs = ['v', 'd', 'j']
if attr in gene_attrs:
xlabels = lcommon.sort_by_gene_number(k2ydata.keys())
rcmp.diff_plot(k2ydata, outbase, xlabels, attr.upper(),
ylabel='Frequency')
else:
rcmp.diff_plot(k2ydata, outbase, label=attr.upper(), xmin=-0.5,
xmax=30.5, ylabel='Frequency')
for attr in objs[0].get_attrs_depth2():
k_k2ydata = model_get_ydata2(objs, attr)
for k, k2ydata in k_k2ydata.iteritems():
if not isinstance(k, str):
k = str(k)
outbase = os.path.join(outdir, "%s_%s" % (attr, k))
if attr in ['v2del', 'j2del', 'd2del']:
rcmp.diff_plot(k2ydata, outbase, label=attr.upper(), xmin=-0.5,
xmax=16.5, ylabel='Frequency')
else:
rcmp.diff_plot(k2ydata, outbase, label=attr.upper(),
ylabel='Frequency')
def draw_gene_usage(name2obj, attr, type, outbase, genes, opts=None):
'''xaxis: genes
yaxis: relative usage
one line/ sample
'''
if opts:
axes, fig, pdf = drawcommon.get_axes(outfile=outbase,
outfmt=opts.plotformat, dpi=opts.dpi)
else:
axes, fig, pdf = drawcommon.get_axes(outfile=outbase)
#genes = sorted(genes)
genes = libcommon.sort_by_gene_number(genes)
xdata = range(len(genes))
group2line = {}
lines = []
linenames = []
for name, obj in name2obj.iteritems():
ydata = gu_get_sample_data(obj, attr, type, genes)
line, = axes.plot(xdata, ydata, color=obj.color, marker=obj.marker,
markeredgecolor=obj.color, linestyle='-')
lines.append(line)
linenames.append(obj.name)
if obj.group not in group2line:
group2line[obj.group] = line
if len(linenames) > 10:
linenames = sorted(group2line.keys())
lines = [group2line[group] for group in linenames]
drawcommon.set_legend(axes, lines, linenames)
drawcommon.adjust_ticklabels(axes, xrotation=75)
drawcommon.set_grid(axes)
drawcommon.edit_spine(axes)
xlabels = [g.lstrip("TRB") for g in genes]
drawcommon.set_xticks(axes, xdata, xlabels)
axes.set_xlim(-0.5, len(xlabels) + 0.5)
axes.set_ylim(bottom=-0.005)
drawcommon.adjust_ticklabels(axes, xrotation=75)
drawcommon.set_labels(axes, xlabel="Gene", ylabel="%% of total %s" % attr)
if opts:
drawcommon.write_image(fig, pdf, opts.plotformat, outbase, opts.dpi)
else:
drawcommon.write_image(fig, pdf, outname=outbase)
def model_diff(indir1, indir2, outdir):
# indir1 contains sam1_model.pickle, sam2_model.pickle ...
# indir2: similarly to indir1
name1 = os.path.basename(indir1.rstrip("/"))
name2 = os.path.basename(indir2.rstrip("/"))
objs1 = lcommon.load_pickledir(indir1)
objs2 = lcommon.load_pickledir(indir2)
rcommon.union_models(objs1 + objs2)
# statistic for each attr for differential usage:
statsdir = os.path.join(outdir, "stat_tests")
system("mkdir -p %s" % statsdir)
model_diff_signi_test(objs1, objs2, statsdir)
# if True: # HACK
# return
med1 = rcommon.model_get_median(objs1)
# med2 = model_get_median(objs2)
for attr in med1.get_attrs_depth1():
outbase1 = os.path.join(outdir, "%s_%s_cmp2_%s" % (attr, name2, name1))
k2diffs1 = diff_sample_model(med1, objs2, attr)
gene_attrs = ["v", "d", "j", "dj"]
if attr in gene_attrs:
if attr == "dj":
xlabels = sorted(k2diffs1.keys())
else:
xlabels = lcommon.sort_by_gene_number(k2diffs1.keys())
diff_plot(k2diffs1, outbase1, xlabels, attr.upper(), ymin=-0.1, ymax=0.1)
else:
diff_plot(k2diffs1, outbase1, label=attr.upper(), xmin=-0.5, xmax=30.5, ymin=-0.02, ymax=0.05)
# outbase2 = os.path.join(outdir, "%s_%s_cmp2_%s" % (attr, name1, name2))
# k2diffs2 = diff_sample_model(med2, objs1, attr)
# diff_plot(k2diffs2, outbase2)
for attr in med1.get_attrs_depth2():
k_k2diffs1 = diff_sample_model2(med1, objs2, attr)
for k, k2diffs1 in k_k2diffs1.iteritems():
if not isinstance(k, str):
k = str(k)
outbase1 = os.path.join(outdir, "%s_%s_%s_cmp2_%s" % (attr, k, name2, name1))
if attr in ["v2del", "j2del", "d2del"]:
diff_plot(k2diffs1, outbase1, label=attr.upper(), xmin=-0.5, xmax=16.5)
else:
diff_plot(k2diffs1, outbase1, label=attr.upper())
def trackclone_latex_tab(f, clone2groups):
sortedclones = libcommon.sort_by_gene_number(clone2groups.keys())
#for clone in sorted(clone2groups.keys()):
for clone in sortedclones:
cloneitems = clone.split('_')
v = cloneitems[0]
seq = cloneitems[1]
j = cloneitems[2]
groupfreqs = clone2groups[clone]
freqstr = ''
for gfreqs in groupfreqs:
if not gfreqs:
freqstr += "& Absent"
else:
freqs = [float(freq) for freq in gfreqs.rstrip(',').split(',')]
sortedfreqs = sorted(freqs, reverse=True)
sortedfreqs = ["%.4f" % (freq * 100) for freq in sortedfreqs]
freqstr += " & %s" % ",".join(sortedfreqs)
f.write("%s & %s & %s %s\\\\\n" % (v, seq, j, freqstr))
f.write("\\hline\n")
def draw_gene_usage_avr(name2obj, attr, type, outbase, genes, opts, bar=False):
'''xaxis: genes
yaxis: relative usage
one line/ sample
'''
axes, fig, pdf = drawcommon.get_axes(outfile=outbase,
outfmt=opts.plotformat, dpi=opts.dpi)
#genes = sorted(genes)
#genes = sorted(genes, key=lambda g: libcommon.get_gene_number(g))
genes = libcommon.sort_by_gene_number(genes)
xdata = range(len(genes))
g2ydata = {}
g2color = {}
for name, obj in name2obj.iteritems():
ydata = gu_get_sample_data(obj, attr, type, genes)
g = obj.group
if g not in g2ydata:
g2ydata[g] = [[y] for y in ydata]
g2color[g] = obj.color
else:
assert len(g2ydata[g]) == len(ydata)
for i, y in enumerate(ydata):
g2ydata[g][i].append(y)
barwidth = (1.0 - 0.35) / len(g2ydata.keys())
lines = []
linenames = []
#for g, ydata in g2ydata.iteritems():
for i, g in enumerate(sorted(g2ydata.keys())):
ydata = g2ydata[g]
ydata = [ylist if ylist else [0.0] for ylist in ydata]
mean_ydata = [np.mean(ylist) for ylist in ydata]
std_ydata = [np.std(ylist) for ylist in ydata]
if not bar:
line, = axes.plot(xdata, mean_ydata, color=g2color[g],
linestyle='-', markeredgecolor=g2color[g],
marker='o')
lines.append(line)
axes.errorbar(xdata, mean_ydata, yerr=std_ydata, color=g2color[g],
linestyle="None", marker="None")
else:
group_xdata = [x + barwidth * i for x in xdata]
line = axes.bar(group_xdata, mean_ydata, barwidth, yerr=std_ydata,
color=g2color[g], ecolor="#424242",
edgecolor=g2color[g])
lines.append(line[0])
linenames.append(g)
drawcommon.set_legend(axes, lines, linenames)
drawcommon.set_grid(axes)
drawcommon.edit_spine(axes)
xlabels = [g.lstrip("TRB") for g in genes]
if not bar:
drawcommon.set_xticks(axes, xdata, xlabels)
else:
xticks = [x + 0.325 for x in xdata]
drawcommon.set_xticks(axes, xticks, xlabels)
axes.set_xlim(-0.5, len(xlabels) + 0.5)
axes.set_ylim(bottom=-0.005)
drawcommon.adjust_ticklabels(axes, xrotation=75)
drawcommon.set_labels(axes, xlabel="Gene", ylabel="%% of total %s" % attr)
drawcommon.write_image(fig, pdf, opts.plotformat, outbase, opts.dpi)
