def test_basic_usage():
'''
Test basic usage -- whether Hist are properly derived from\
boost-histogram.
'''
# Test normal Hist
h = Hist(axis.Regular(10, 0, 1, name='x'))
h.fill([0.35, 0.35, 0.45])
assert h[2] == 0
assert h[3] == 2
assert h[4] == 1
assert h[5] == 0
assert h[{0:2}] == 0
assert h[{0:3}] == 2
assert h[{0:4}] == 1
assert h[{0:5}] == 0
# Test multi-axis Hist
h = Hist(
axis.Regular(10, 0, 1, name="x"),
axis.Regular(10, 0, 1, name="y"),
axis.Integer(0, 2, name="z")
)
h.fill([0.35, 0.35, 0.35, 0.45, 0.55, 0.55, 0.55],
[0.35, 0.35, 0.45, 0.45, 0.45, 0.45, 0.45],
[0, 0, 1, 1, 1, 1, 1])
def get_mute_hist(self, mtype):
if self.args.mutate_from_scratch:
mean_mute_val = self.args.scratch_mute_freq
if self.args.same_mute_freq_for_all_seqs:
hist = Hist(1, mean_mute_val - utils.eps,
mean_mute_val + utils.eps)
hist.fill(mean_mute_val)
else:
n_entries = 500
length_vals = [
v
for v in numpy.random.exponential(mean_mute_val, n_entries)
] # count doesn't work on numpy.ndarray objects
max_val = 0.8 # this is arbitrary, but you shouldn't be calling this with anything that gets a significant number anywhere near there, anyway
if length_vals.count(max_val):
print '%s lots of really high mutation rates treegenerator::get_mute_hist()' % utils.color(
'yellow', 'warning')
length_vals = [min(v, max_val) for v in length_vals]
hist = Hist(30, 0., max_val)
for val in length_vals:
hist.fill(val)
hist.normalize()
else:
hist = Hist(fname=self.parameter_dir + '/' + mtype +
'-mean-mute-freqs.csv')
return hist
def addplot(oindexlist, ofracslist, n_seqs, fname, title):
hist = Hist(30, 0., 1.)
for ofracs in ofracslist:
hist.fill(ofracs)
fig, ax = self.plotting.mpl_init()
hist.mpl_plot(ax, remove_empty_bins=True)
ax.text(0.65,
0.8 * ax.get_ylim()[1],
'size: %d' % n_seqs,
fontsize=20,
fontweight='bold')
ax.text(0.65,
0.7 * ax.get_ylim()[1],
'h: %.2f' %
utils.fay_wu_h(line=None,
restrict_to_region=restrict_to_region,
occurence_indices=oindexlist,
n_seqs=n_seqs),
fontsize=20,
fontweight='bold')
regionstr = restrict_to_region + ' ' if restrict_to_region is not None else ''
self.plotting.mpl_finish(
ax,
plotdir,
fname,
title=title,
xlabel=regionstr + 'mutation frequency',
ylabel=regionstr + 'density of mutations',
xticks=[0, 1],
log=''
) # xticks=[min(occurence_fractions), max(occurence_fractions)],
self.addfname(fnames, fname)
def test_basic_usage():
h = Hist(axis.Regular(10, 0, 1))
h.fill([0.35, 0.35, 0.45])
assert h[2] == 0
assert h[3] == 2
assert h[4] == 1
assert h[5] == 0
def get_cluster_size_hist(partition, rebin=None):
sizes = [len(c) for c in partition]
nbins = max(sizes)
# if nbins > 30:
# rebin = 2
if rebin is not None:
nbins = int(float(nbins) / rebin)
hist = Hist(nbins, 0.5, max(sizes) + 0.5)
for sz in sizes:
hist.fill(sz)
return hist
def get_cluster_size_hist(partition, rebin=None):
sizes = [len(c) for c in partition]
nbins = max(sizes)
# if nbins > 30:
# rebin = 2
if rebin is not None:
nbins = int(float(nbins) / rebin)
hist = Hist(nbins, 0.5, max(sizes) + 0.5)
for sz in sizes:
hist.fill(sz)
return hist
def test_basic_usage():
# Check hist with only one axis
h = Hist(axis.Regular(10, 0, 1))
h.fill([0.35, 0.35, 0.45])
assert h[2] == 0
assert h[3] == 2
assert h[4] == 1
assert h[5] == 0
# Check hist with two axes
h = Hist(axis.Regular(10, 0, 1), axis.Regular(10, 0, 1))
h.fill([0.35, 0.35, 0.45], [0.65, 0.65, 0.85])
assert h[3, 6] == 2
assert h[4, 8] == 1
assert h[3, 5] == 0
# Checking hist with axis type bool
h = Hist(axis.bool())
h.fill([0, 1, 1])
assert h[0] == 1
assert h[1] == 2
# check if there are exactly two bins (accessing h[2] raises IndexError)
with pytest.raises(IndexError):
assert h[2] == 0
# check if flow is disabled (if view() with or without flow gives the same output)
assert (h.view() == h.view(flow=True)).all()
h = Hist(axis.Regular(10, 0, 1), axis.Regular(10, 0, 1))
h.fill([0.35, 0.35, 0.45], [0.65, 0.65, 0.85])
# Check indexing using dict and bh.loc()
h2 = h[loc(0.35), :]
# Broken in 0.6.2, fixed now
h3 = h[{0: loc(0.35)}]
assert (h2.view() == h3.view()).all()
def make_hist():
ax, fill = request.param
h = Hist(ax)
if fill is int:
h.fill(np.random.randn(10))
elif fill is bool:
h.fill(np.random.randint(0, 1, size=10) == 1)
elif fill is str:
h.fill(np.random.choice(("T", "F"), size=10))
return h
def test_plot1d_auto_handling():
"""
Test plot() by comparing against a reference image generated via
`pytest --mpl-generate-path=tests/baseline`
"""
np.random.seed(42)
h = Hist(
axis.Regular(10, 0, 10, name="variable", label="variable"),
axis.StrCategory("", name="dataset", growth=True),
)
h_nameless = Hist(
axis.Regular(10, 0, 10),
axis.StrCategory("", growth=True),
)
h.fill(dataset="A", variable=np.random.normal(3, 2, 100))
h.fill(dataset="B", variable=np.random.normal(5, 2, 100))
h.fill(dataset="C", variable=np.random.normal(7, 2, 100))
h_nameless.fill(np.random.normal(3, 2, 1000), "A")
h_nameless.fill(np.random.normal(5, 2, 1000), "B")
h_nameless.fill(np.random.normal(7, 2, 1000), "C")
fig, (ax1, ax2) = plt.subplots(2, 2, figsize=(14, 10))
assert h.plot(ax=ax1[0])
assert h_nameless.plot(ax=ax2[0])
# Discrete axis plotting not yet implemented
# assert h.plot(ax=ax1[1], overlay='variable')
# assert h.plot(ax=ax2[1], overlay=1)
return fig
def get_cluster_size_hist(partition):
sizes = [len(c) for c in partition]
hist = Hist(max(sizes), 0.5, max(sizes) + 0.5)
for sz in sizes:
hist.fill(sz)
return hist
def test_basic_usage():
# Check if axis without name raises an error
with pytest.raises(KeyError):
h_named = NamedHist(
axis.Regular(10, 0, 1, name="x"),
axis.Regular(10, 0, 1)
)
h_named = NamedHist(
axis.Regular(10, 0, 1, name="x"),
axis.Regular(10, 0, 1, name="y")
) # NamedHist should require axis.Regular to have a name set
# Check if filling without keyword raises error
with pytest.raises(ValueError):
h_named.fill([0.35, 0.35, 0.45], y=[0.65, 0.75, 0.85])
h_named.fill(x=[0.35, 0.35, 0.45], y=[5, 10, 14]) # Fill should be keyword only, with the names
h_normal = Hist(
axis.Regular(10, 0, 1, name="x"),
axis.Regular(10, 5, 15, name="y")
)
h_normal.fill([0.35, 0.35, 0.45], [0.65, 0.75, 0.85])
assert (h_named.view() == h_normal.view()).all()
h = NamedHist(
axis.Regular(10, 0, 1, name='x')
)
h.fill(x=[0.35, 0.35, 0.45])
# Example of a test that should be made to pass:
assert h[{'x': 2}] == 0 # Should work
assert h[{'x': 3}] == 2 # Should work
assert h[{'x': 4}] == 1 # Should work
assert h[{'x': 5}] == 0 # Should work
# Additional Test cases on indexing
h2 = h_normal[{0: slice(1, 5, None), 1: slice(None, 5, None)}]
h3 = h_named[{'y': slice(None, 5, None), 'x': slice(1, 5, None)}]
# Check if indexing by axis name works correctly
assert (h2.view() == h3.view()).all()
h2 = h_normal[{0: 3}]
h3 = h_named[{'x': 3}]
# Check if indexing works correctly
assert (h2.view() == h3.view()).all()
h2 = h_normal[{0: loc(0.35)}]
h3 = h_normal[loc(0.35), :]
h4 = h_named[{'x': loc(0.35)}]
# Checking if indexing with loc() works correctly
assert (h2.view() == h3.view()).all()
assert (h3.view() == h4.view()).all()
h2 = h_normal[{1: slice(None, None, sum)}]
h3 = h_named[{'y': slice(None, None, sum)}]
# Check if indexing with sum works correctly
assert (h2.view() == h3.view()).all()
biggest_adiffs = sorted(chfo,
key=lambda q: chfo[q]['max_abs_diff'],
reverse=True)
for uid in biggest_adiffs[:5]:
print '%-3d %6.3f' % (chfo[uid]['imax'], chfo[uid]['max_abs_diff'])
utils.print_reco_event(annotations[uid])
n_above_cutoff = len(
[_ for cfo in chfo.values() if cfo['max_abs_diff'] > args.cutoff])
chimeric_fraction = n_above_cutoff / float(len(chfo))
print ' %d / %d = %.3f above chimeric cutoff' % (n_above_cutoff, len(chfo),
chimeric_fraction)
hmaxval = Hist(45, 0., 0.65)
for uid in annotations:
hmaxval.fill(chfo[uid]['max_abs_diff'])
himax = Hist(75, 0., 400)
for uid in annotations:
himax.fill(chfo[uid]['imax'])
utils.prep_dir(args.plotdir, wildlings=['*.svg', '*.csv'])
import matplotlib
from matplotlib import pyplot as plt
fig, ax = plotting.mpl_init()
xvals, yvals = zip(*[(v['imax'], v['max_abs_diff']) for v in chfo.values()])
plt.scatter(xvals, yvals, alpha=0.4)
print 'writing to %s' % args.plotdir
plotting.mpl_finish(ax,
args.plotdir,
def make_single_joyplot(self,
sorted_clusters,
annotations,
repertoire_size,
plotdir,
plotname,
plot_high_mutation=False,
cluster_indices=None,
title=None,
debug=False):
def gety(minval, maxval, xmax, x):
slope = (maxval - minval) / xmax
return slope * x + minval
def getnmutelist(cluster):
return annotations[':'.join(cluster)]['n_mutations']
colors = ['#006600', '#3399ff', '#ffa500']
# goldenrod '#daa520'
# red '#cc0000',
# dark red '#990012'
# purple '#a821c7'
# grey '#808080'
dpi = 80
xpixels = 450
ypixels = max(400, 10 * len(sorted_clusters))
fig, ax = self.plotting.mpl_init(figsize=(xpixels / dpi,
ypixels / dpi))
min_linewidth = 0.3
max_linewidth = 12
# min_alpha = 0.1
# max_alpha = 1.
# linewidth = 7
alpha = 0.55
ymin, ymax = 9999, 0
iclust_global = 0 # index within this plot
yticks, yticklabels = [], []
high_mutation_clusters = []
biggest_n_mutations = None
if debug:
print ' %s %d x %d %s' % (plotname, xpixels, ypixels,
utils.color('red', 'high mutation')
if plot_high_mutation else '')
print ' size frac yval median mean'
for csize, cluster_group in itertools.groupby(sorted_clusters,
key=lambda c: len(c)):
cluster_group = sorted(list(cluster_group),
key=lambda c: numpy.median(getnmutelist(c)))
n_clusters = len(cluster_group)
repfracstr = self.get_repfracstr(csize, repertoire_size)
for iclust in range(len(
cluster_group)): # index within the clusters of this size
cluster = cluster_group[iclust]
nmutelist = sorted(getnmutelist(cluster))
nmedian = numpy.median(nmutelist)
nmean = numpy.mean(
nmutelist) # maybe should use this instead of the median?
if biggest_n_mutations is None or nmutelist[
-1] > biggest_n_mutations:
biggest_n_mutations = nmutelist[-1]
if nmedian > self.n_max_mutations and not plot_high_mutation:
high_mutation_clusters.append(cluster)
continue
yval = len(sorted_clusters) - iclust_global
if yval < ymin:
ymin = yval
if yval > ymax:
ymax = yval
yticks.append(yval)
# yticklabels.append('%d' % csize)
yticklabels.append(repfracstr)
base_color = colors[iclust_global % len(colors)]
qti_n_muted = {}
if self.args.queries_to_include is not None:
queries_to_include_in_this_cluster = set(cluster) & set(
self.args.queries_to_include)
if len(queries_to_include_in_this_cluster) > 0:
unsorted_nmutelist = getnmutelist(cluster)
qti_n_muted = {
uid: unsorted_nmutelist[cluster.index(uid)]
for uid in queries_to_include_in_this_cluster
} # add a red line for each of 'em (i.e. color that hist bin red)
if plot_high_mutation:
xtext = 1.1
elif float(nmedian) / self.n_max_mutations < 0.5:
xtext = 0.75
else:
xtext = 0.1
ax.text(xtext * self.n_max_mutations,
yval,
' '.join(
sorted(queries_to_include_in_this_cluster,
key=lambda q: qti_n_muted[q])),
color='red',
fontsize=8)
if debug:
print ' %5s %-10s %4.1f %6.1f %6.1f' % (
'%d' % csize if iclust == 0 else '', repfracstr
if iclust == 0 else '', yval, nmedian, nmean)
nbins = nmutelist[-1] - nmutelist[0] + 1
hist = Hist(nbins, nmutelist[0] - 0.5, nmutelist[-1] + 0.5)
for nm in nmutelist:
hist.fill(nm)
assert hist.overflow_contents() == 0. # includes underflows
xmax = max(
hist.bin_contents) # NOTE no relation to <ymax> above
for ibin in range(1, hist.n_bins + 1):
linewidth = gety(min_linewidth, max_linewidth, xmax,
hist.bin_contents[ibin])
color = base_color
# alpha = gety(min_alpha, max_alpha, xmax, hist.bin_contents[ibin])
for nmuted in qti_n_muted.values():
if hist.find_bin(nmuted) == ibin:
color = 'red'
if hist.bin_contents[ibin] == 0.:
color = 'grey'
linewidth = min_linewidth
alpha = 0.4
ax.plot([hist.low_edges[ibin], hist.low_edges[ibin + 1]],
[yval, yval],
color=color,
linewidth=linewidth,
alpha=alpha,
solid_capstyle='butt')
if cluster_indices is not None:
xtext = nmutelist[
-1] if plot_high_mutation else self.n_max_mutations # NOTE reuse of <xtext> (arg)
xwidth = ax.get_xlim()[1] - ax.get_xlim(
)[0] if plot_high_mutation else self.n_max_mutations
ax.text(0.05 * xwidth + xtext,
yval,
str(cluster_indices[':'.join(cluster)]),
color=base_color,
fontsize=6,
alpha=alpha,
fontdict={'weight': 'bold'})
ax.text(0.12 * xwidth + xtext,
yval,
str(csize),
color=base_color,
fontsize=6,
alpha=alpha,
fontdict={'weight': 'bold'})
iclust_global += 1
xbounds = [-0.2, self.n_max_mutations] if not plot_high_mutation else [
self.n_max_mutations, biggest_n_mutations
]
ybounds = [0.95 * ymin, 1.05 * ymax]
n_ticks = 5
if len(yticks) > n_ticks:
yticks = [
yticks[i] for i in range(0, len(yticks),
int(len(yticks) / float(n_ticks - 1)))
]
yticklabels = [
yticklabels[i]
for i in range(0, len(yticklabels),
int(len(yticklabels) / float(n_ticks - 1)))
]
self.plotting.mpl_finish(
ax,
plotdir,
plotname,
xlabel='N mutations',
ylabel='fraction of repertoire',
title=title, # ylabel = 'clonal family size'
xbounds=xbounds,
ybounds=ybounds,
yticks=yticks,
yticklabels=yticklabels,
adjust={'left': 0.25})
return high_mutation_clusters
label=label_gen,
name=voi_gen,
underflow=True,
overflow=True),
hist.axis.Variable(voi_v.bins,
label=label_rec,
name=voi_rec,
underflow=True,
overflow=True),
storage=hist.storage.Weight(), # like ROOT's Sumw2()
)
gen_array = arrays[voi_gen]
rec_array = arrays[voi_rec]
hist2d.fill(gen_array, rec_array, weight=arrays[branch_weight])
outputDir = os.path.join(
os.environ.get('CMSSW_BASE'),
'src/UHH2/HighPtSingleTop/output/Analysis/mainsel', year,
channel, 'unfolding/migration_matrices/')
os.system('mkdir -p ' + outputDir)
outfilePath = os.path.join(
outputDir, 'migration_matrix__' + args.drds + '__' +
voi_k + '__' + gen_level + '__region_' + region + '__' +
year + '_' + channel + '.root')
with uproot.recreate(outfilePath) as outfile:
outfile['migration_matrix'] = hist2d
outfile['axis_label_gen'] = label_gen
outfile['axis_label_rec'] = label_rec
def get_cluster_size_hist(partition):
sizes = [len(c) for c in partition]
hist = Hist(max(sizes), 0.5, max(sizes) + 0.5)
for sz in sizes:
hist.fill(sz)
return hist
def gk(uids):
return ':'.join(uids)
glfo = glutils.read_glfo(args.infile.replace('.csv', '-glfo'), locus='igh')
annotations = {}
with open(args.infile) as csvfile:
reader = csv.DictReader(csvfile)
for line in reader:
if line['v_gene'] == '': # failed (i.e. couldn't find an annotation)
continue
utils.process_input_line(line) # converts strings in the csv file to floats/ints/dicts/etc.
utils.add_implicit_info(glfo, line) # add stuff to <line> that's useful, isn't written to the csv since it's redundant
annotations[gk(line['unique_ids'])] = line
chfo = {uid : utils.get_chimera_max_abs_diff(annotations[uid], iseq=0) for uid in annotations}
biggest_adiffs = sorted(chfo, key=lambda q: chfo[q][1], reverse=True)
for uid in biggest_adiffs[:10]:
print chfo[uid]
utils.print_reco_event(annotations[uid])
htmp = Hist(45, 0., 0.65)
for uid in annotations:
htmp.fill(chfo[uid][1])
utils.prep_dir(args.plotdir, wildlings=['*.svg', '*.csv'])
plotname = 'mfreq-diff'
plotting.draw_no_root(htmp, plotdir=args.plotdir, plotname=plotname, shift_overflows=True, xtitle='abs mfreq diff', ytitle='seqs')
plotting.draw_no_root(htmp, plotdir=args.plotdir, plotname=plotname + '-log', shift_overflows=True, log='y', xtitle='abs mfreq diff', ytitle='seqs')
print 'writing to %s' % args.plotdir
htmp.write('%s/%s.csv' % (args.plotdir, plotname))
def make_single_size_vs_shm_plot(self,
sorted_clusters,
annotations,
repertoire_size,
base_plotdir,
plotname,
n_max_mutations=100,
plot_high_mutation=False,
title=None,
debug=False):
import plotting
def gety(minval, maxval, xmax, x):
slope = (maxval - minval) / xmax
return slope * x + minval
def getnmutelist(cluster):
return annotations[':'.join(cluster)]['n_mutations']
colors = ['#006600', '#3399ff', '#ffa500']
# goldenrod '#daa520'
# red '#cc0000',
# dark red '#990012'
# purple '#a821c7'
# grey '#808080'
dpi = 80
xpixels = 450
ypixels = max(400, 10 * len(sorted_clusters))
fig, ax = plotting.mpl_init(figsize=(xpixels / dpi, ypixels / dpi))
min_linewidth = 0.3
max_linewidth = 12
# min_alpha = 0.1
# max_alpha = 1.
# linewidth = 7
alpha = 0.55
ymin, ymax = 9999, 0
iclust_global = 0
yticks, yticklabels = [], []
high_mutation_clusters = []
biggest_n_mutations = None
if debug:
print ' %s %d x %d' % (
plotname, xpixels, ypixels
) #, utils.color('red', 'high mutation') if plot_high_mutation else '')
print ' size frac yval median mean'
for csize, cluster_group in itertools.groupby(sorted_clusters,
key=lambda c: len(c)):
cluster_group = sorted(list(cluster_group),
key=lambda c: numpy.median(getnmutelist(c)))
n_clusters = len(cluster_group)
repfracstr = self.get_repfracstr(csize, repertoire_size)
for iclust in range(len(cluster_group)):
cluster = cluster_group[iclust]
nmutelist = sorted(getnmutelist(cluster))
nmedian = numpy.median(nmutelist)
nmean = numpy.mean(
nmutelist) # maybe should use this instead of the median?
if biggest_n_mutations is None or nmutelist[
-1] > biggest_n_mutations:
biggest_n_mutations = nmutelist[-1]
yval = len(sorted_clusters) - iclust_global
if yval < ymin:
ymin = yval
if yval > ymax:
ymax = yval
yticks.append(yval)
yticklabels.append('%d' % csize)
# yticklabels.append(repfracstr)
base_color = colors[iclust_global % len(colors)]
if self.args.queries_to_include is not None:
queries_to_include_in_this_cluster = set(cluster) & set(
self.args.queries_to_include)
if len(queries_to_include_in_this_cluster) > 0:
base_color = 'red'
if plot_high_mutation:
xtext = 1.1
elif float(nmedian) / n_max_mutations < 0.5:
xtext = 0.75
else:
xtext = 0.1
ax.text(xtext * n_max_mutations,
yval,
' '.join(queries_to_include_in_this_cluster),
color='red',
fontsize=8)
if debug:
print ' %5s %-10s %4.1f %6.1f %6.1f' % (
'%d' % csize if iclust == 0 else '', repfracstr
if iclust == 0 else '', yval, nmedian, nmean),
if nmedian > n_max_mutations and not plot_high_mutation:
if debug:
print '%s' % utils.color('red', 'high mutation')
high_mutation_clusters.append(cluster)
continue
if debug:
print ''
nbins = nmutelist[-1] - nmutelist[0] + 1
hist = Hist(nbins, nmutelist[0] - 0.5, nmutelist[-1] + 0.5)
for nm in nmutelist:
hist.fill(nm)
assert hist.overflow_contents() == 0. # includes underflows
xmax = max(hist.bin_contents) # float(csize)
for ibin in range(1, hist.n_bins + 1):
linewidth = gety(min_linewidth, max_linewidth, xmax,
hist.bin_contents[ibin])
color = base_color
# alpha = gety(min_alpha, max_alpha, xmax, hist.bin_contents[ibin])
if hist.bin_contents[ibin] == 0.:
color = 'grey'
linewidth = min_linewidth
alpha = 0.4
ax.plot([hist.low_edges[ibin], hist.low_edges[ibin + 1]],
[yval, yval],
color=color,
linewidth=linewidth,
alpha=alpha,
solid_capstyle='butt')
iclust_global += 1
xbounds = [-0.2, n_max_mutations] if not plot_high_mutation else [
n_max_mutations, biggest_n_mutations
]
ybounds = [0.95 * ymin, 1.05 * ymax]
n_ticks = 5
if len(yticks) > n_ticks:
yticks = [
yticks[i] for i in range(0, len(yticks),
int(len(yticks) / float(n_ticks - 1)))
]
yticklabels = [
yticklabels[i]
for i in range(0, len(yticklabels),
int(len(yticklabels) / float(n_ticks - 1)))
]
plotting.mpl_finish(ax,
base_plotdir + '/overall',
plotname,
xlabel='N mutations',
ylabel='clonal family size',
title=title,
xbounds=xbounds,
ybounds=ybounds,
yticks=yticks,
yticklabels=yticklabels,
adjust={'left': 0.18})
return high_mutation_clusters