def build_inferred_seq(self, seq, all_germlines, outline):
assert self.excisions[0]['region'] == 'v' # makes it easier a.t.m.
assert self.excisions[1]['region'] == 'j'
assert self.excisions[2]['region'] == 'd'
germlines, hmms, ihmms = {}, {}, {}
for region in utils.regions:
germlines[region] = all_germlines[region][utils.unsanitize_name(self.best_matches[region]['target_name'])]
hmms[region] = self.best_matches[region]['hmm_seq']
ihmms[region] = germlines[region].find(hmms[region].upper()) # position at which the consensus (hmm) starts in the germline sequence
try:
assert ihmms[region] >= 0
except:
print germlines[region]
print hmms[region].upper()
print ihmms[region]
assert False
print ' hmm for %s runs from %d to %d (inclusive)' % (region, ihmms[region], ihmms[region] + len(hmms[region]) - 1)
outline['v_5p_del'] = ihmms['v'] # TODO kinda otter be zero
outline['v_3p_del'] = len(germlines['v']) - ihmms['v'] - len(hmms['v']) # len(germlines['v']) - len(hmms['v']) - germlines['v'].find(hmms['v'].upper())
outline['d_5p_del'] = ihmms['d'] # germlines['d'].find(hmms['d'].upper())
outline['d_3p_del'] = len(germlines['d']) - ihmms['d'] - len(hmms['d']) # len(germlines['d']) - len(hmms['d']) - germlines['d'].find(hmms['d'].upper())
outline['j_5p_del'] = ihmms['j'] # germlines['j'].find(hmms['j'].upper())
outline['j_3p_del'] = len(germlines['j']) - ihmms['j'] - len(hmms['j']) # TODO kinda otter be zero
for ex in self.excisions:
match = self.best_matches[ex['region']]
print ' excised match %s: %d --> %d' % (ex['region'], ex['from'], ex['to'])
print ' test %s' % match['test_seq']
hmm_start = ihmms[ex['region']]
hmm_end = ihmms[ex['region']] + len(hmms[ex['region']]) - 1
print ' hmm %s' % (hmm_start * '.' + match['hmm_seq'].upper() + (len(germlines[ex['region']]) - ihmms[ex['region']] - len(hmms[ex['region']])) * '.') # NOTE ali_from includes the d part!
print ' germline %s' % all_germlines[ex['region']][utils.unsanitize_name(match['target_name'])]
#----------------------------------------------------------------------------------------
# NOTE these are inclusive
seq_match_start, seq_match_end = {}, {}
seq_match_start['v'] = self.best_matches['v']['ali_from'] - 1
seq_match_end['v'] = seq_match_start['v'] + len(hmms['v']) - 1
seq_match_start['d'] = self.excisions[0]['to'] - self.excisions[0]['from'] + self.best_matches['d']['ali_from']
seq_match_end['d'] = seq_match_start['d'] + len(hmms['d']) - 1
seq_match_start['j'] = self.excisions[0]['to'] - self.excisions[0]['from'] + self.best_matches['j']['ali_from']
seq_match_end['j'] = seq_match_start['j'] + len(hmms['j']) - 1
outline['vd_insertion'] = seq[seq_match_end['v']+1 : seq_match_start['d']]
outline['dj_insertion'] = seq[seq_match_end['d']+1 : seq_match_start['j']]
actual_seq_length = len(seq)
inferred_seq_length = outline['v_5p_del'] + len(hmms['v']) + len(outline['vd_insertion']) + len(hmms['d']) + len(outline['dj_insertion']) + len(hmms['j']) + outline['j_3p_del']
print ' actual %d inferred %d' % (actual_seq_length,inferred_seq_length)
if actual_seq_length != inferred_seq_length:
outline['ack'] = True
def make_transition_plot(self, gene_name, model):
""" NOTE shares a lot with make_mutefreq_plot() in python/paramutils.py """
fig, ax = plotting.mpl_init()
fig.set_size_inches(plotting.plot_ratios[utils.get_region(gene_name)])
ibin = 0
print utils.color_gene(utils.unsanitize_name(gene_name))
legend_colors = set() # add a color to this the first time you plot it
for state in model.states:
# bin label
ax.text(-0.5 + ibin, -0.075, paramutils.simplify_state_name(state.name), rotation='vertical', size=8)
sorted_to_states = {}
for name in state.transitions.keys():
if name.find('IG') == 0 or name.find('TR') == 0:
sorted_to_states[name] = int(paramutils.simplify_state_name(name))
else:
sorted_to_states[name] = name
sorted_to_states = sorted(sorted_to_states.items(), key=operator.itemgetter(1))
total = 0.0
for to_state, simple_to_state in sorted_to_states:
prob = state.transitions[to_state]
alpha = 0.6
width = 3
if 'insert' in str(simple_to_state):
label = 'insert'
color = '#3498db' # blue
elif str(simple_to_state) == 'end':
label = 'end'
color = 'red'
else: # regional/internal states
assert to_state.find('IG') == 0 or to_state.find('TR') == 0
label = 'internal'
color = 'green'
label_to_use = None
if color not in legend_colors:
label_to_use = label
legend_colors.add(color)
# horizontal line at height total+prob
ax.plot([-0.5 + ibin, 0.5 + ibin], [total + prob, total + prob], color=color, linewidth=width, alpha=alpha, label=label_to_use)
# vertical line from total to total + prob
ax.plot([ibin, ibin], [total + 0.01, total + prob], color=color, alpha=alpha, linewidth=width)
midpoint = 0.5*(prob + 2*total)
# ax.text(ibin, midpoint, paramutils.simplify_state_name(to_state)) # nicely labels the midpoint of the chunk between lines, but there isn't really room for it
total += prob
ibin += 1
ax.get_xaxis().set_visible(False)
plotting.mpl_finish(ax, self.base_plotdir + '/transitions', gene_name, ybounds=(-0.01, 1.01), xbounds=(-3, len(model.states) + 3), leg_loc=(0.95, 0.1), adjust={'left' : 0.1, 'right' : 0.8}, leg_prop={'size' : 8})
def make_transition_plot(self, gene_name, model):
""" NOTE shares a lot with make_mutefreq_plot() in python/paramutils.py """
fig, ax = plotting.mpl_init()
fig.set_size_inches(plotting.plot_ratios[utils.get_region(gene_name)])
ibin = 0
print utils.color_gene(utils.unsanitize_name(gene_name))
legend_colors = set() # add a color to this the first time you plot it
for state in model.states:
# bin label
ax.text(-0.5 + ibin, -0.075, paramutils.simplify_state_name(state.name), rotation='vertical', size=8)
sorted_to_states = {}
for name in state.transitions.keys():
if name.find('IG') == 0:
sorted_to_states[name] = int(paramutils.simplify_state_name(name))
else:
sorted_to_states[name] = name
sorted_to_states = sorted(sorted_to_states.items(), key=operator.itemgetter(1))
total = 0.0
for to_state, simple_to_state in sorted_to_states:
prob = state.transitions[to_state]
alpha = 0.6
width = 3
if 'insert' in str(simple_to_state):
label = 'insert'
color = '#3498db' # blue
elif str(simple_to_state) == 'end':
label = 'end'
color = 'red'
else: # regional/internal states
assert to_state.find('IG') == 0
label = 'internal'
color = 'green'
label_to_use = None
if color not in legend_colors:
label_to_use = label
legend_colors.add(color)
# horizontal line at height total+prob
ax.plot([-0.5 + ibin, 0.5 + ibin], [total + prob, total + prob], color=color, linewidth=width, alpha=alpha, label=label_to_use)
# vertical line from total to total + prob
ax.plot([ibin, ibin], [total + 0.01, total + prob], color=color, alpha=alpha, linewidth=width)
midpoint = 0.5*(prob + 2*total)
# ax.text(ibin, midpoint, paramutils.simplify_state_name(to_state)) # nicely labels the midpoint of the chunk between lines, but there isn't really room for it
total += prob
ibin += 1
ax.get_xaxis().set_visible(False)
plotting.mpl_finish(ax, self.base_plotdir + '/transitions', gene_name, ybounds=(-0.01, 1.01), xbounds=(-3, len(model.states) + 3), leg_loc=(0.95, 0.1), adjust={'left' : 0.1, 'right' : 0.8}, leg_prop={'size' : 8})
def make_mutefreq_plot(plotdir, gene_name, positions):
nuke_colors = {'A':kRed+1, 'C':kBlue-7, 'G':kOrange-3, 'T':kGreen+2}
ibin = 0
drawn_name_texts, lines, vlines, texts = {}, {}, {}, {}
for info in positions:
posname = info['name']
# make label below bin
drawn_name_texts[posname] = TPaveText(-0.5 + ibin, -0.1, 0.5 + ibin, -0.05)
drawn_name_texts[posname].SetBorderSize(0)
drawn_name_texts[posname].SetFillColor(0)
drawn_name_texts[posname].SetFillStyle(0)
drawn_name_texts[posname].AddText(-0.5 + ibin, -0.075, simplify_state_name(posname))
total = 0.0
lines[posname], vlines[posname], texts[posname] = [], [], []
for nuke, prob in sorted(info['nuke_freqs'].items(), key=operator.itemgetter(1), reverse=True):
# horizontal line at height total+prob
lines[posname].append(TLine(-0.5 + ibin, total + prob, 0.5 + ibin, total + prob))
lines[posname][-1].SetLineWidth(6)
# vertical line from total to total+prob
vlines[posname].append(TLine(ibin, total, ibin, total + prob))
vlines[posname][-1].SetLineWidth(6)
vlines[posname][-1].SetLineColor(nuke_colors[nuke])
# write [ACGT] at midpoint between total and total+prob
midpoint = 0.5*(prob + 2*total)
texts[posname].append(TPaveText(-0.5 + ibin, midpoint-0.04, 0.5 + ibin, midpoint + 0.01))
texts[posname][-1].AddText(-0.5 + ibin, midpoint, nuke)
texts[posname][-1].SetBorderSize(0)
texts[posname][-1].SetFillColor(0)
texts[posname][-1].SetFillStyle(0)
total += prob
ibin += 1
cvn = TCanvas('cvn-2', '', 1000, 300)
n_bins = ibin
hframe = TH1D(gene_name + '-emission-frame', utils.unsanitize_name(gene_name), n_bins, -0.5, n_bins - 0.5)
hframe.SetNdivisions(202, 'y')
hframe.SetNdivisions(0, 'x')
hframe.Draw()
for state_name in lines.keys():
drawn_name_texts[state_name].Draw()
for itrans in range(len(lines[state_name])):
# lines[state_name][itrans].Draw() # hm, maybe don't need the horizontal lines any more
vlines[state_name][itrans].Draw()
# texts[state_name][itrans].Draw() # don't label the bases at the moment, you can tell by the color just fine
cvn.SaveAs(plotdir + '/plots/' + gene_name + '.png')
def make_mutefreq_plot(plotdir, gene_name, positions):
import plotting
""" NOTE shares a lot with make_transition_plot() in bin/plot-hmms.py. """
nuke_colors = {'A' : 'red', 'C' : 'blue', 'G' : 'orange', 'T' : 'green'}
fig, ax = plotting.mpl_init()
fig.set_size_inches(plotting.plot_ratios[utils.get_region(gene_name)])
ibin = 0
print utils.color_gene(utils.unsanitize_name(gene_name))
legend_colors = set()
for info in positions:
posname = info['name']
# make label below bin
ax.text(-0.5 + ibin, -0.075, simplify_state_name(posname), rotation='vertical', size=8)
total = 0.0
alpha = 0.6
for nuke, prob in sorted(info['nuke_freqs'].items(), key=operator.itemgetter(1), reverse=True):
color = nuke_colors[nuke]
label_to_use = None
if color not in legend_colors:
label_to_use = nuke
legend_colors.add(color)
# horizontal line at height total+prob
ax.plot([-0.5 + ibin, 0.5 + ibin], [total + prob, total + prob], color=color, alpha=alpha, linewidth=3, label=label_to_use)
# vertical line from total to total + prob
ax.plot([ibin, ibin], [total + 0.01, total + prob], color=color, alpha=alpha, linewidth=3)
# # write [ACGT] at midpoint between total and total+prob
# midpoint = 0.5*(prob + 2*total)
# ... *redacted*
total += prob
ibin += 1
ax.get_xaxis().set_visible(False)
plotting.mpl_finish(ax, plotdir, gene_name, ybounds=(-0.01, 1.01), xbounds=(-3, len(positions) + 3), leg_loc=(0.95, 0.1), adjust={'left' : 0.1, 'right' : 0.8}, leg_prop={'size' : 8})
def make_mutefreq_plot(plotdir, gene_name, positions):
""" NOTE shares a lot with make_transition_plot() in bin/plot-hmms.py. """
nuke_colors = {'A' : 'red', 'C' : 'blue', 'G' : 'orange', 'T' : 'green'}
fig, ax = plotting.mpl_init()
fig.set_size_inches(plotting.plot_ratios[utils.get_region(gene_name)])
ibin = 0
print utils.color_gene(utils.unsanitize_name(gene_name))
legend_colors = set()
for info in positions:
posname = info['name']
# make label below bin
ax.text(-0.5 + ibin, -0.075, simplify_state_name(posname), rotation='vertical', size=8)
total = 0.0
alpha = 0.6
for nuke, prob in sorted(info['nuke_freqs'].items(), key=operator.itemgetter(1), reverse=True):
color = nuke_colors[nuke]
label_to_use = None
if color not in legend_colors:
label_to_use = nuke
legend_colors.add(color)
# horizontal line at height total+prob
ax.plot([-0.5 + ibin, 0.5 + ibin], [total + prob, total + prob], color=color, alpha=alpha, linewidth=3, label=label_to_use)
# vertical line from total to total + prob
ax.plot([ibin, ibin], [total + 0.01, total + prob], color=color, alpha=alpha, linewidth=3)
# # write [ACGT] at midpoint between total and total+prob
# midpoint = 0.5*(prob + 2*total)
# ... *redacted*
total += prob
ibin += 1
ax.get_xaxis().set_visible(False)
plotting.mpl_finish(ax, plotdir, gene_name, ybounds=(-0.01, 1.01), xbounds=(-3, len(positions) + 3), leg_loc=(0.95, 0.1), adjust={'left' : 0.1, 'right' : 0.8}, leg_prop={'size' : 8})
def plot_single_variable(args, varname, hlist, outdir, pathnameclues):
if varname in plotconfig.gene_usage_columns:
hlist = plotting.add_bin_labels_not_in_all_hists(hlist)
no_labels = False
xline, bounds, figsize = None, None, None
stats = args.extra_stats
translegend = [0.0, -0.2]
xtitle, ytitle = hlist[0].xtitle, hlist[0].ytitle
if xtitle == '': # arg, plotting.py thinks default should be None, hist.py thinks it's ''
xtitle = None
if '-mean-bins' in varname:
raise Exception(
'darn, I was hoping I wasn\'t making these plots any more')
plottitle = plotconfig.plot_titles[
varname] if varname in plotconfig.plot_titles else varname
ytitle = 'frequency' if args.normalize else 'counts'
if 'mute-freqs/v' in pathnameclues or 'mute-freqs/d' in pathnameclues or 'mute-freqs/j' in pathnameclues:
assert not args.normalize
ytitle = 'mutation freq'
if varname in plotconfig.gene_usage_columns:
xtitle = 'allele'
if hlist[0].n_bins == 2:
stats = ' 0-bin' # print the fraction of entries in the zero bin into the legend (i.e. the fraction correct)
# elif hlist[0].bin_labels.count('') == hlist[0].n_bins + 2:
# xtitle = '???'
line_width_override = None
if args.performance_plots:
if 'hamming_to_true_naive' in varname:
xtitle = 'hamming distance'
if '_normed' in varname:
xtitle = 'fractional ' + xtitle
elif '_vs_mute_freq' in varname:
xtitle = 'mutation freq'
ytitle = 'fraction correct'
if varname[0] == 'v' or varname[0] == 'j':
translegend = [-0.4, -0.4]
elif varname.find('_gene') == 1:
xtitle = ''
ytitle = 'fraction correct'
else:
xtitle = 'inferred - true'
bounds = plotconfig.true_vs_inferred_hard_bounds.setdefault(
varname, None)
else:
bounds = plotconfig.default_hard_bounds.setdefault(varname, None)
if bounds is None and 'insertion' in varname:
bounds = plotconfig.default_hard_bounds.setdefault(
'all_insertions', None)
if varname in plotconfig.gene_usage_columns:
no_labels = True
if 'j_' not in varname:
figsize = (10, 5)
line_width_override = 1
elif 'per-gene-per-position/v' in pathnameclues:
figsize = (20, 5)
bounds = plotconfig.default_hard_bounds.setdefault(
utils.unsanitize_name(varname), None)
if 'IG' in varname or 'TR' in varname:
if 'mute-freqs' in pathnameclues:
gene = utils.unsanitize_name(varname)
plottitle = gene # + ' -- mutation frequency'
xtitle = 'position'
if utils.get_region(gene) == 'j':
translegend = [0.1, 0.] #(-0.35, -0.02)
else:
translegend = [0.15, -0.02]
xline = None
if args.glfo is not None:
if utils.get_region(gene) in utils.conserved_codons[
args.locus]:
xline = args.glfo[utils.conserved_codons[args.locus][
utils.get_region(gene)] + '-positions'][gene]
else:
ilastdash = varname.rfind('-')
gene = utils.unsanitize_name(varname[:ilastdash])
base_varname = varname[ilastdash + 1:]
base_plottitle = plotconfig.plot_titles[
base_varname] if base_varname in plotconfig.plot_titles else ''
plottitle = gene + ' -- ' + base_plottitle
if len(hlist) > 9: # skootch it down so they (maybe) all fit
translegend[1] -= 0.5
if args.translegend is not None: # override with the command line
translegend = args.translegend
if args.extra_stats == 'auto': # kind of hackey
if xtitle == 'inferred - true':
stats = 'absmean'
else:
stats = 'mean'
# draw that little #$*(!
linewidths = [
line_width_override,
] if line_width_override is not None else args.linewidths
alphas = [0.6 for _ in range(len(hlist))]
plotting.draw_no_root(
hlist[0],
plotname=varname,
plotdir=outdir,
more_hists=hlist[1:],
write_csv=False,
stats=stats,
bounds=bounds,
shift_overflows=(os.path.basename(outdir) != 'gene-call'),
plottitle=plottitle,
colors=args.colors,
xtitle=xtitle,
ytitle=ytitle,
xline=xline,
normalize=(args.normalize and '_vs_mute_freq' not in varname),
linewidths=linewidths,
alphas=alphas,
errors=True,
figsize=figsize,
no_labels=no_labels,
log=args.log,
translegend=translegend)
def make_transition_plot(self, gene_name, model):
ibin = 0
drawn_name_texts, lines, texts = {}, {}, {}
for state in model.states:
if utils.get_region(gene_name) in self.skip_boring_states:
if state.name != 'init' and len(state.transitions) == 1: # skip uninteresting states
to_state = state.transitions.keys()[0] # skip states with only transitions to end
if to_state == 'end':
continue
if find_state_number(state.name) + 1 == find_state_number(to_state): # skip states with only transitions to next state
continue
drawn_name_texts[state.name] = TPaveText(-0.5 + ibin, -0.1, 0.5 + ibin, -0.05)
drawn_name_texts[state.name].SetBorderSize(0)
drawn_name_texts[state.name].SetFillColor(0)
drawn_name_texts[state.name].SetFillStyle(0)
drawn_name_texts[state.name].AddText(-0.5 + ibin, -0.075, paramutils.simplify_state_name(state.name))
sorted_to_states = {}
for name in state.transitions.keys():
if name.find('IGH') == 0:
sorted_to_states[name] = int(paramutils.simplify_state_name(name))
else:
sorted_to_states[name] = name
sorted_to_states = sorted(sorted_to_states.items(), key=operator.itemgetter(1))
total = 0.0
lines[state.name], texts[state.name] = [], []
for to_state, simple_to_state in sorted_to_states:
prob = state.transitions[to_state]
lines[state.name].append(TLine(-0.5 + ibin, total + prob, 0.5 + ibin, total + prob))
lines[state.name][-1].SetLineColor(kGreen+2)
lines[state.name][-1].SetLineWidth(6)
midpoint = 0.5*(prob + 2*total)
texts[state.name].append(TPaveText(-0.5 + ibin, midpoint-0.04, 0.5 + ibin, midpoint + 0.01))
texts[state.name][-1].AddText(-0.5 + ibin, midpoint, paramutils.simplify_state_name(to_state))
texts[state.name][-1].SetBorderSize(0)
texts[state.name][-1].SetFillColor(0)
texts[state.name][-1].SetFillStyle(0)
total += prob
ibin += 1
cvn = TCanvas('mod-cvn', '', 1000, 400)
n_bins = ibin
hframe = TH1D(model.name + '-transition-frame', utils.unsanitize_name(model.name), n_bins, -0.5, n_bins - 0.5)
if utils.get_region(gene_name) in self.skip_boring_states:
hframe.SetTitle(hframe.GetTitle() + ' (skipped boring states)')
hframe.SetNdivisions(202, 'y')
hframe.SetNdivisions(0, 'x')
hframe.Draw()
for state_name in lines.keys():
drawn_name_texts[state_name].Draw()
for itrans in range(len(lines[state_name])):
lines[state_name][itrans].Draw()
texts[state_name][itrans].Draw()
cvn.SaveAs(self.base_plotdir + '/transitions/plots/' + gene_name + '.png')
def make_transition_plot(self, gene_name, model):
ibin = 0
drawn_name_texts, lines, texts = {}, {}, {}
for state in model.states:
if utils.get_region(gene_name) in self.skip_boring_states:
if state.name != 'init' and len(
state.transitions) == 1: # skip uninteresting states
to_state = state.transitions.keys()[
0] # skip states with only transitions to end
if to_state == 'end':
continue
if find_state_number(state.name) + 1 == find_state_number(
to_state
): # skip states with only transitions to next state
continue
drawn_name_texts[state.name] = TPaveText(-0.5 + ibin, -0.1,
0.5 + ibin, -0.05)
drawn_name_texts[state.name].SetBorderSize(0)
drawn_name_texts[state.name].SetFillColor(0)
drawn_name_texts[state.name].SetFillStyle(0)
drawn_name_texts[state.name].AddText(
-0.5 + ibin, -0.075,
paramutils.simplify_state_name(state.name))
sorted_to_states = {}
for name in state.transitions.keys():
if name.find('IGH') == 0:
sorted_to_states[name] = int(
paramutils.simplify_state_name(name))
else:
sorted_to_states[name] = name
sorted_to_states = sorted(sorted_to_states.items(),
key=operator.itemgetter(1))
total = 0.0
lines[state.name], texts[state.name] = [], []
for to_state, simple_to_state in sorted_to_states:
prob = state.transitions[to_state]
lines[state.name].append(
TLine(-0.5 + ibin, total + prob, 0.5 + ibin, total + prob))
lines[state.name][-1].SetLineColor(kGreen + 2)
lines[state.name][-1].SetLineWidth(6)
midpoint = 0.5 * (prob + 2 * total)
texts[state.name].append(
TPaveText(-0.5 + ibin, midpoint - 0.04, 0.5 + ibin,
midpoint + 0.01))
texts[state.name][-1].AddText(
-0.5 + ibin, midpoint,
paramutils.simplify_state_name(to_state))
texts[state.name][-1].SetBorderSize(0)
texts[state.name][-1].SetFillColor(0)
texts[state.name][-1].SetFillStyle(0)
total += prob
ibin += 1
cvn = TCanvas('mod-cvn', '', 1000, 400)
n_bins = ibin
hframe = TH1D(model.name + '-transition-frame',
utils.unsanitize_name(model.name), n_bins, -0.5,
n_bins - 0.5)
if utils.get_region(gene_name) in self.skip_boring_states:
hframe.SetTitle(hframe.GetTitle() + ' (skipped boring states)')
hframe.SetNdivisions(202, 'y')
hframe.SetNdivisions(0, 'x')
hframe.Draw()
for state_name in lines.keys():
drawn_name_texts[state_name].Draw()
for itrans in range(len(lines[state_name])):
lines[state_name][itrans].Draw()
texts[state_name][itrans].Draw()
cvn.SaveAs(self.base_plotdir + '/transitions/plots/' + gene_name +
'.png')
def compare_directories(args, xtitle='', use_hard_bounds=''):
"""
Read all the histograms stored as .csv files in <args.plotdirs>, and overlay them on a new plot.
If there's a <varname> that's missing from any dir, we skip that plot entirely and print a warning message.
"""
# print 'TODO move csvs to a subdir not named "plots"'
# utils.prep_dir(args.outdir + '/plots', multilings=['*.png', '*.svg', '*.csv'])
utils.prep_dir(args.outdir, multilings=['*.png', '*.svg', '*.csv'])
if args.leaves_per_tree is not None:
assert len(args.leaves_per_tree) == len(args.plotdirs)
# read hists from <args.plotdirs>
hists = []
for idir in range(len(args.plotdirs)):
string_to_ignore = None if args.strings_to_ignore is None else args.strings_to_ignore[idir]
hist_list = get_hists_from_dir(args.plotdirs[idir], args.names[idir], string_to_ignore=string_to_ignore)
hists.append(hist_list)
# then loop over all the <varname>s we found
all_names, all_means, all_sems, all_normalized_means = [], [], [], []
# ----------------------------------------------------------------------------------------
# vs_rebin = 2
vs_rebin = 1
if 'v_gene_right_vs_mute_freq' in hists[0].keys():
add_gene_calls_vs_mute_freq_plots(args, hists, rebin=vs_rebin)
# ----------------------------------------------------------------------------------------
for varname, hist in hists[0].items():
# add the hists
all_hists = [hist,]
missing_hist = False
for idir in range(1, len(args.plotdirs)):
try: # add the hist
all_hists.append(hists[idir][varname])
except KeyError: # oops, didn't find it in this dir, so skip this variable entirely
print args.names[idir], varname
all_hists.append(Hist(1, 0, 1))
if '_gene' in varname and '_vs_' not in varname: # for the gene usage frequencies we need to make sure all the plots have the genes in the same order
all_hists = add_bin_labels_not_in_all_hists(all_hists)
if args.calculate_mean_info:
raise Exception('needs updating (at least to remove plots/ )')
meaninfo = get_mean_info(all_hists)
all_names.append(varname)
all_means.append(meaninfo['means'])
all_sems.append(meaninfo['sems'])
all_normalized_means.append(meaninfo['normalized_means'])
meaninfo['mean_bin_hist'].write(args.outdir + '/plots/' + varname + '-mean-bins.csv')
# bullshit complicated config stuff
bounds, no_labels, figsize = None, False, None
translegend = (0.0, -0.2)
extrastats, log = '', ''
xtitle, ytitle, xline, normalization_bounds = hist.xtitle, hist.ytitle, None, None
simplevarname = varname.replace('-mean-bins', '')
plottitle = plotconfig.plot_titles[simplevarname] if simplevarname in plotconfig.plot_titles else simplevarname
if args.normalize:
ytitle = 'frequency'
if 'mute-freqs/v' in args.plotdirs[0] or 'mute-freqs/d' in args.plotdirs[0] or 'mute-freqs/j' in args.plotdirs[0]:
assert not args.normalize
ytitle = 'mutation freq'
if '_gene' in varname and '_vs_' not in varname:
xtitle = 'allele'
if hist.n_bins == 2:
extrastats = ' 0-bin' # print the fraction of entries in the zero bin into the legend (i.e. the fraction correct)
else:
xtitle = 'bases'
line_width_override = None
rebin = args.rebin
errors = not args.no_errors
if args.plot_performance:
if 'hamming_to_true_naive' in varname:
xtitle = 'hamming distance'
if '_normed' in varname:
xtitle = 'fractional ' + xtitle
elif '_vs_mute_freq' in varname:
xtitle = 'mutation freq'
ytitle = 'fraction correct'
if varname[0] == 'v' or varname[0] == 'j':
translegend = (-0.4, -0.4)
# errors = True
rebin = vs_rebin
else:
xtitle = 'inferred - true'
bounds = plotconfig.true_vs_inferred_hard_bounds.setdefault(varname, None)
else:
bounds = plotconfig.default_hard_bounds.setdefault(varname.replace('-mean-bins', ''), None)
if bounds is None and 'insertion' in varname:
bounds = plotconfig.default_hard_bounds.setdefault('all_insertions', None)
if '_gene' in varname and '_vs_' not in varname:
no_labels = True
if 'j_' not in varname:
figsize = (10, 5)
line_width_override = 1
elif 'mute-freqs/v' in args.plotdirs[0] or 'mute-freqs/j' in args.plotdirs[0]:
figsize = (10, 5)
bounds = plotconfig.default_hard_bounds.setdefault(utils.unsanitize_name(varname.replace('-mean-bins', '')), None)
if 'IGH' in varname:
if 'mute-freqs' in args.plotdirs[0]:
gene = utils.unsanitize_name(simplevarname)
plottitle = gene # + ' -- mutation frequency'
xtitle = 'position'
if utils.get_region(gene) == 'j':
translegend = (0.1, 0.) #(-0.35, -0.02)
else:
translegend = (0.15, -0.02)
xline = None
if utils.get_region(gene) == 'v' and args.cyst_positions is not None:
xline = args.cyst_positions[gene]
# normalization_bounds = (int(cyst_positions[gene]) - 70, None)
elif utils.get_region(gene) == 'j' and args.tryp_positions is not None:
xline = args.tryp_positions[gene]
# normalization_bounds = (None, int(tryp_positions[gene]) + 5)
else:
ilastdash = simplevarname.rfind('-')
gene = utils.unsanitize_name(simplevarname[:ilastdash])
base_varname = simplevarname[ilastdash + 1 :]
base_plottitle = plotconfig.plot_titles[base_varname] if base_varname in plotconfig.plot_titles else ''
plottitle = gene + ' -- ' + base_plottitle
# draw that little #$*(!
linewidths = [line_width_override, ] if line_width_override is not None else args.linewidths
assert args.leaves_per_tree is None
# scale_errors = math.sqrt(args.leaves_per_tree[idir]) if args.leaves_per_tree is not None else args.scale_errors
draw_no_root(all_hists[0], plotname=varname, plotdir=args.outdir, more_hists=all_hists[1:], write_csv=False, stats=args.stats + ' ' + extrastats, bounds=bounds,
shift_overflows=False, errors=errors, scale_errors=args.scale_errors, rebin=rebin, plottitle=plottitle, colors=args.colors, linestyles=args.linestyles,
xtitle=xtitle, ytitle=ytitle, xline=xline, normalize=(args.normalize and '_vs_mute_freq' not in varname),
linewidths=linewidths, markersizes=args.markersizes, figsize=figsize, no_labels=no_labels, log=log, translegend=translegend, alphas=args.alphas)
if args.calculate_mean_info:
assert False
# write mean info
with opener('w')(args.outdir + '/plots/means.csv') as meanfile:
writer = csv.DictWriter(meanfile, ('name', 'means', 'sems', 'normalized-means'))
writer.writeheader()
for ivar in range(len(all_means)):
writer.writerow({
'name':all_names[ivar],
'means':':'.join([str(m) for m in all_means[ivar]]),
'sems':':'.join([str(s) for s in all_sems[ivar]]),
'normalized-means':':'.join([str(nm) for nm in all_normalized_means[ivar]])
})
if not args.only_csv_plots:
make_html(args.outdir)
def compare_directories(args, xtitle='', use_hard_bounds=''):
"""
Read all the histograms stored as .csv files in <args.plotdirs>, and overlay them on a new plot.
If there's a <varname> that's missing from any dir, we skip that plot entirely and print a warning message.
"""
utils.prep_dir(args.outdir + '/plots',
multilings=['*.png', '*.svg', '*.csv'])
if args.leaves_per_tree is not None:
assert len(args.leaves_per_tree) == len(args.plotdirs)
# read hists from <args.plotdirs>
hists = []
for idir in range(len(args.plotdirs)):
string_to_ignore = None if args.strings_to_ignore is None else args.strings_to_ignore[
idir]
hists.append(
get_hists_from_dir(args.plotdirs[idir] + '/plots',
args.names[idir],
string_to_ignore=string_to_ignore))
# then loop over all the <varname>s we found
all_names, all_means, all_sems, all_normalized_means = [], [], [], []
# ----------------------------------------------------------------------------------------
vs_rebin = 2
if 'v_gene_right_vs_mute_freq' in hists[0].keys():
add_gene_calls_vs_mute_freq_plots(args, hists, rebin=vs_rebin)
# ----------------------------------------------------------------------------------------
for varname, hist in hists[0].items():
# add the hists
all_hists = [
hist,
]
missing_hist = False
for idir in range(1, len(args.plotdirs)):
try: # add the hist
all_hists.append(hists[idir][varname])
except KeyError: # oops, didn't find it in this dir, so skip this variable entirely
print args.names[idir], varname
all_hists.append(TH1D())
if '_gene' in varname and '_vs_' not in varname: # for the gene usage frequencies we need to make sure all the plots have the genes in the same order
all_hists = add_bin_labels_not_in_all_hists(all_hists)
if not args.dont_calculate_mean_info:
meaninfo = get_mean_info(all_hists)
all_names.append(varname)
all_means.append(meaninfo['means'])
all_sems.append(meaninfo['sems'])
all_normalized_means.append(meaninfo['normalized_means'])
meaninfo['mean_bin_hist'].write(args.outdir + '/plots/' + varname +
'-mean-bins.csv')
# bullshit complicated config stuff
var_type = 'int' if hist.GetXaxis().GetBinLabel(1) == '' else 'bool'
bounds, cwidth, cheight, translegend, no_labels = None, None, None, (
0.0, 0.0), False
extrastats, log = '', ''
xtitle, ytitle, xline, draw_str, normalization_bounds = hist.GetXaxis(
).GetTitle(), hist.GetYaxis().GetTitle(), None, None, None
simplevarname = varname.replace('-mean-bins', '')
plottitle = plotconfig.plot_titles[
simplevarname] if simplevarname in plotconfig.plot_titles else simplevarname
if args.normalize:
ytitle = 'frequency'
if 'mute-freqs/v' in args.plotdirs[
0] or 'mute-freqs/d' in args.plotdirs[
0] or 'mute-freqs/j' in args.plotdirs[0]:
assert not args.normalize
ytitle = 'mutation freq'
args.graphify = True
if '_gene' in varname and '_vs_' not in varname:
xtitle = 'allele'
gStyle.SetNdivisions(0, "x")
# gStyle.SetLabelSize(0.00010, 'X')
if hist.GetNbinsX() == 2:
extrastats = ' 0-bin' # print the fraction of entries in the zero bin into the legend (i.e. the fraction correct)
if 'v_gene' in varname:
pass
# log += 'y'
else:
gStyle.SetNdivisions(505, "x")
xtitle = 'bases'
line_width_override = None
rebin = args.rebin
errors = not args.no_errors
if args.plot_performance:
if 'hamming_to_true_naive' in varname:
xtitle = 'hamming distance'
if '_normed' in varname:
xtitle = 'fractional ' + xtitle
elif '_vs_mute_freq' in varname:
xtitle = 'mutation freq'
ytitle = 'fraction correct'
if varname[0] == 'v' or varname[0] == 'j':
translegend = (-0.4, -0.4)
# errors = True
rebin = vs_rebin
else:
xtitle = 'inferred - true'
bounds = plotconfig.true_vs_inferred_hard_bounds.setdefault(
varname, None)
else:
bounds = plotconfig.default_hard_bounds.setdefault(
varname.replace('-mean-bins', ''), None)
if bounds is None and 'insertion' in varname:
bounds = plotconfig.default_hard_bounds.setdefault(
'all_insertions', None)
if '_gene' in varname and '_vs_' not in varname:
no_labels = True
if 'j_' not in varname:
cwidth, cheight = 1000, 500
line_width_override = 1
elif 'mute-freqs/v' in args.plotdirs[
0] or 'mute-freqs/j' in args.plotdirs[0]:
cwidth, cheight = 1000, 500
bounds = plotconfig.default_hard_bounds.setdefault(
utils.unsanitize_name(varname.replace('-mean-bins', '')),
None)
if 'IGH' in varname:
if 'mute-freqs' in args.plotdirs[0]:
gene = utils.unsanitize_name(simplevarname)
plottitle = gene # + ' -- mutation frequency'
xtitle = 'position'
if utils.get_region(gene) == 'j':
translegend = (0.1, 0.) #(-0.35, -0.02)
else:
translegend = (0.15, -0.02)
xline = None
if utils.get_region(
gene) == 'v' and args.cyst_positions is not None:
xline = args.cyst_positions[gene]['cysteine-position']
# normalization_bounds = (int(cyst_positions[gene]['cysteine-position']) - 70, None)
elif utils.get_region(
gene) == 'j' and args.tryp_positions is not None:
xline = int(args.tryp_positions[gene])
# normalization_bounds = (None, int(tryp_positions[gene]) + 5)
else:
ilastdash = simplevarname.rfind('-')
gene = utils.unsanitize_name(simplevarname[:ilastdash])
base_varname = simplevarname[ilastdash + 1:]
base_plottitle = plotconfig.plot_titles[
base_varname] if base_varname in plotconfig.plot_titles else ''
plottitle = gene + ' -- ' + base_plottitle
# draw that little #$*(!
linewidths = [
line_width_override,
] if line_width_override is not None else args.linewidths
assert args.leaves_per_tree is None
# scale_errors = math.sqrt(args.leaves_per_tree[idir]) if args.leaves_per_tree is not None else args.scale_errors
draw(all_hists[0],
var_type,
plotname=varname,
plotdir=args.outdir,
more_hists=all_hists[1:],
write_csv=False,
stats=args.stats + ' ' + extrastats,
bounds=bounds,
shift_overflows=False,
errors=errors,
scale_errors=args.scale_errors,
rebin=rebin,
plottitle=plottitle,
colors=args.colors,
linestyles=args.linestyles,
xtitle=xtitle,
ytitle=ytitle,
xline=xline,
draw_str=draw_str,
normalize=(args.normalize and '_vs_mute_freq' not in varname),
normalization_bounds=normalization_bounds,
linewidths=linewidths,
markersizes=args.markersizes,
cwidth=cwidth,
cheight=cheight,
no_labels=no_labels,
graphify=args.graphify,
log=log,
translegend=translegend)
if not args.dont_calculate_mean_info:
# write mean info
with opener('w')(args.outdir + '/plots/means.csv') as meanfile:
writer = csv.DictWriter(
meanfile, ('name', 'means', 'sems', 'normalized-means'))
writer.writeheader()
for ivar in range(len(all_means)):
writer.writerow({
'name':
all_names[ivar],
'means':
':'.join([str(m) for m in all_means[ivar]]),
'sems':
':'.join([str(s) for s in all_sems[ivar]]),
'normalized-means':
':'.join([str(nm) for nm in all_normalized_means[ivar]])
})
check_call(
['./bin/permissify-www', args.outdir]
) # NOTE this should really permissify starting a few directories higher up
check_call(['./bin/makeHtml', args.outdir, '3', 'null', 'svg'])
def make_mutefreq_plot(plotdir, gene_name, positions, debug=False):
import plotting
""" NOTE shares a lot with make_transition_plot() in bin/plot-hmms.py. """
nuke_colors = {'A': 'red', 'C': 'blue', 'G': 'orange', 'T': 'green'}
fig, ax = plotting.mpl_init()
fig.set_size_inches(plotting.plot_ratios[utils.get_region(gene_name)])
ibin = 0
if debug:
print ' %s' % utils.color_gene(utils.unsanitize_name(gene_name))
legend_colors = set()
for info in positions:
posname = info['name']
# make label below bin for position and germline nuke
ax.text(-0.5 + ibin,
-0.075,
simplify_state_name(posname),
rotation='vertical',
size=8)
ax.text(-0.5 + ibin,
-0.15,
info.get('gl_nuke', '?'),
fontsize=10,
fontweight='bold')
sorted_nukes, _ = zip(*sorted(info['nuke_freqs'].items(),
key=operator.itemgetter(1),
reverse=True))
if 'gl_nuke' in info and info['gl_nuke'] in info[
'nuke_freqs']: # put the germline nuke first if we have it (second clause is for states with germline N))
sorted_nukes = [info['gl_nuke']] + [
n for n in sorted_nukes if n != info['gl_nuke']
]
total = 0.0
alpha = 0.6
for nuke in sorted_nukes:
prob = info['nuke_freqs'][nuke]
color = nuke_colors[nuke]
label_to_use = None
if color not in legend_colors:
label_to_use = nuke
legend_colors.add(color)
# horizontal line at height total+prob
ax.plot([-0.5 + ibin, 0.5 + ibin], [total + prob, total + prob],
color=color,
alpha=alpha,
linewidth=3,
label=label_to_use)
# vertical line from total to total + prob
ax.plot([ibin, ibin], [total + 0.01, total + prob],
color=color,
alpha=alpha,
linewidth=3)
# # write [ACGT] at midpoint between total and total+prob
# midpoint = 0.5*(prob + 2*total)
# ... *redacted*
total += prob
ibin += 1
ax.get_xaxis().set_visible(False)
plotting.mpl_finish(ax,
plotdir,
gene_name,
ybounds=(-0.01, 1.01),
xbounds=(-3, len(positions) + 3),
leg_loc=(0.95, 0.1),
adjust={
'left': 0.1,
'right': 0.8
},
leg_prop={'size': 8})
def plot_single_variable(args, varname, hlist, outdir, pathnameclues):
if varname in plotconfig.gene_usage_columns:
hlist = plotting.add_bin_labels_not_in_all_hists(hlist)
no_labels = False
xline, bounds, figsize = None, None, None
translegend = (0.0, -0.2)
extrastats, log = '', ''
xtitle, ytitle = hlist[0].xtitle, hlist[0].ytitle
if xtitle == '': # arg, plotting.py thinks default should be None, hist.py thinks it's ''
xtitle = None
if '-mean-bins' in varname:
raise Exception('darn, I was hoping I wasn\'t making these plots any more')
plottitle = plotconfig.plot_titles[varname] if varname in plotconfig.plot_titles else varname
ytitle = 'frequency' if args.normalize else 'counts'
if 'mute-freqs/v' in pathnameclues or 'mute-freqs/d' in pathnameclues or 'mute-freqs/j' in pathnameclues:
assert not args.normalize
ytitle = 'mutation freq'
if varname in plotconfig.gene_usage_columns:
xtitle = 'allele'
if hlist[0].n_bins == 2:
extrastats = ' 0-bin' # print the fraction of entries in the zero bin into the legend (i.e. the fraction correct)
# elif hlist[0].bin_labels.count('') == hlist[0].n_bins + 2:
# xtitle = '???'
line_width_override = None
if args.performance_plots:
if 'hamming_to_true_naive' in varname:
xtitle = 'hamming distance'
if '_normed' in varname:
xtitle = 'fractional ' + xtitle
elif '_vs_mute_freq' in varname:
xtitle = 'mutation freq'
ytitle = 'fraction correct'
if varname[0] == 'v' or varname[0] == 'j':
translegend = (-0.4, -0.4)
elif varname.find('_gene') == 1:
xtitle = ''
ytitle = 'fraction correct'
else:
xtitle = 'inferred - true'
bounds = plotconfig.true_vs_inferred_hard_bounds.setdefault(varname, None)
else:
bounds = plotconfig.default_hard_bounds.setdefault(varname, None)
if bounds is None and 'insertion' in varname:
bounds = plotconfig.default_hard_bounds.setdefault('all_insertions', None)
if varname in plotconfig.gene_usage_columns:
no_labels = True
if 'j_' not in varname:
figsize = (10, 5)
line_width_override = 1
elif 'per-gene-per-position/v' in pathnameclues:
figsize = (20, 5)
bounds = plotconfig.default_hard_bounds.setdefault(utils.unsanitize_name(varname), None)
if 'IG' in varname:
if 'mute-freqs' in pathnameclues:
gene = utils.unsanitize_name(varname)
plottitle = gene # + ' -- mutation frequency'
xtitle = 'position'
if utils.get_region(gene) == 'j':
translegend = (0.1, 0.) #(-0.35, -0.02)
else:
translegend = (0.15, -0.02)
xline = None
if args.glfo is not None:
if utils.get_region(gene) in utils.conserved_codons[args.chain]:
xline = args.glfo[utils.conserved_codons[args.chain][utils.get_region(gene)] + '-positions'][gene]
else:
ilastdash = varname.rfind('-')
gene = utils.unsanitize_name(varname[:ilastdash])
base_varname = varname[ilastdash + 1 :]
base_plottitle = plotconfig.plot_titles[base_varname] if base_varname in plotconfig.plot_titles else ''
plottitle = gene + ' -- ' + base_plottitle
# draw that little #$*(!
linewidths = [line_width_override, ] if line_width_override is not None else args.linewidths
alphas = [0.6 for _ in range(len(hlist))]
plotting.draw_no_root(hlist[0], plotname=varname, plotdir=outdir, more_hists=hlist[1:], write_csv=False, stats=extrastats, bounds=bounds,
shift_overflows=(os.path.basename(outdir) != 'gene-call'), plottitle=plottitle, colors=args.colors,
xtitle=xtitle, ytitle=ytitle, xline=xline, normalize=(args.normalize and '_vs_mute_freq' not in varname),
linewidths=linewidths, alphas=alphas, errors=True,
figsize=figsize, no_labels=no_labels, log=log, translegend=translegend)
def make_mutefreq_plot(plotdir, gene_name, positions):
nuke_colors = {
'A': kRed + 1,
'C': kBlue - 7,
'G': kOrange - 3,
'T': kGreen + 2
}
ibin = 0
drawn_name_texts, lines, vlines, texts = {}, {}, {}, {}
for info in positions:
posname = info['name']
# make label below bin
drawn_name_texts[posname] = TPaveText(-0.5 + ibin, -0.1, 0.5 + ibin,
-0.05)
drawn_name_texts[posname].SetBorderSize(0)
drawn_name_texts[posname].SetFillColor(0)
drawn_name_texts[posname].SetFillStyle(0)
drawn_name_texts[posname].AddText(-0.5 + ibin, -0.075,
simplify_state_name(posname))
total = 0.0
lines[posname], vlines[posname], texts[posname] = [], [], []
for nuke, prob in sorted(info['nuke_freqs'].items(),
key=operator.itemgetter(1),
reverse=True):
# horizontal line at height total+prob
lines[posname].append(
TLine(-0.5 + ibin, total + prob, 0.5 + ibin, total + prob))
lines[posname][-1].SetLineWidth(6)
# vertical line from total to total+prob
vlines[posname].append(TLine(ibin, total, ibin, total + prob))
vlines[posname][-1].SetLineWidth(6)
vlines[posname][-1].SetLineColor(nuke_colors[nuke])
# write [ACGT] at midpoint between total and total+prob
midpoint = 0.5 * (prob + 2 * total)
texts[posname].append(
TPaveText(-0.5 + ibin, midpoint - 0.04, 0.5 + ibin,
midpoint + 0.01))
texts[posname][-1].AddText(-0.5 + ibin, midpoint, nuke)
texts[posname][-1].SetBorderSize(0)
texts[posname][-1].SetFillColor(0)
texts[posname][-1].SetFillStyle(0)
total += prob
ibin += 1
cvn = TCanvas('cvn-2', '', 1000, 300)
n_bins = ibin
hframe = TH1D(gene_name + '-emission-frame',
utils.unsanitize_name(gene_name), n_bins, -0.5, n_bins - 0.5)
hframe.SetNdivisions(202, 'y')
hframe.SetNdivisions(0, 'x')
hframe.Draw()
for state_name in lines.keys():
drawn_name_texts[state_name].Draw()
for itrans in range(len(lines[state_name])):
# lines[state_name][itrans].Draw() # hm, maybe don't need the horizontal lines any more
vlines[state_name][itrans].Draw()
# texts[state_name][itrans].Draw() # don't label the bases at the moment, you can tell by the color just fine
cvn.SaveAs(plotdir + '/plots/' + gene_name + '.png')
with opener('r')(infname) as infile:
germlines = utils.read_germlines('../../../recombinator')
reader = csv.DictReader(infile)
for inline in reader:
print 'searching'
# inline['seq'] = inline['seq'][-130:]
searcher = Searcher(inline['seq'], debug=True, n_matches_max=2)
searcher.search()
inferred_group_str = ''
true_group_str = ''
outline = {}
outline['seq'] = inline['seq']
print 'RESULT ',
for region in utils.regions:
inferred_name = searcher.get_best_match_name(region)
outline[region + '_gene'] = utils.unsanitize_name(inferred_name)
true_name = utils.sanitize_name(inline[region + '_gene'])
inferred_group_str += inferred_name
true_group_str += true_name
if inferred_name == 'none':
print ' none',
elif inferred_name == true_name:
print ' - ',
else:
print ' x ',
for region in utils.regions:
print '%3d' % searcher.n_tries[region],
print ''
print ' true'
utils.print_reco_event(germlines, inline, -1, -1)
def compare_directories(args, xtitle='', use_hard_bounds=''):
"""
Read all the histograms stored as .csv files in <args.plotdirs>, and overlay them on a new plot.
If there's a <varname> that's missing from any dir, we skip that plot entirely and print a warning message.
"""
utils.prep_dir(args.outdir + '/plots', multilings=['*.png', '*.svg', '*.csv'])
if args.leaves_per_tree is not None:
assert len(args.leaves_per_tree) == len(args.plotdirs)
# read hists from <args.plotdirs>
hists = []
for idir in range(len(args.plotdirs)):
string_to_ignore = None if args.strings_to_ignore is None else args.strings_to_ignore[idir]
hists.append(get_hists_from_dir(args.plotdirs[idir] + '/plots', args.names[idir], string_to_ignore=string_to_ignore))
# then loop over all the <varname>s we found
all_names, all_means, all_sems, all_normalized_means = [], [], [], []
for varname, hist in hists[0].iteritems():
# add the hists
all_hists = [hist,]
missing_hist = False
for idir in range(1, len(args.plotdirs)):
try: # add the hist
all_hists.append(hists[idir][varname])
except KeyError: # oops, didn't find it in this dir, so skip this variable entirely
print args.names[idir], varname
all_hists.append(TH1D())
if '_gene' in varname: # for the gene usage frequencies we need to make sure all the plots have the genes in the same order
all_hists = add_bin_labels_not_in_all_hists(all_hists)
if not args.dont_calculate_mean_info:
meaninfo = get_mean_info(all_hists)
all_names.append(varname)
all_means.append(meaninfo['means'])
all_sems.append(meaninfo['sems'])
all_normalized_means.append(meaninfo['normalized_means'])
meaninfo['mean_bin_hist'].write(args.outdir + '/plots/' + varname + '-mean-bins.csv')
# bullshit complicated config stuff
var_type = 'int' if hist.GetXaxis().GetBinLabel(1) == '' else 'bool'
bounds, cwidth, cheight, translegend, no_labels = None, None, None, (0.0, 0.0), False
extrastats, log = '', ''
xtitle, ytitle, xline, draw_str, normalization_bounds = hist.GetXaxis().GetTitle(), hist.GetYaxis().GetTitle(), None, None, None
simplevarname = varname.replace('-mean-bins', '')
plottitle = plotconfig.plot_titles[simplevarname] if simplevarname in plotconfig.plot_titles else simplevarname
if args.normalize:
ytitle = 'frequency'
if 'mute-freqs/v' in args.plotdirs[0] or 'mute-freqs/d' in args.plotdirs[0] or 'mute-freqs/j' in args.plotdirs[0]:
assert not args.normalize
ytitle = 'mutation freq'
args.graphify = True
if '_gene' in varname:
xtitle = 'allele'
gStyle.SetNdivisions(0,"x")
# gStyle.SetLabelSize(0.00010, 'X')
if hist.GetNbinsX() == 2:
extrastats = ' 0-bin' # print the fraction of entries in the zero bin into the legend (i.e. the fraction correct)
if 'v_gene' in varname:
pass
# log += 'y'
else:
gStyle.SetNdivisions(505,"x")
xtitle = 'bases'
line_width_override = None
if args.plot_performance:
if 'hamming_to_true_naive' in varname:
xtitle = 'hamming distance'
if '_normed' in varname:
xtitle = 'fractional ' + xtitle
else:
xtitle = 'inferred - true'
bounds = plotconfig.true_vs_inferred_hard_bounds.setdefault(varname, None)
else:
bounds = plotconfig.default_hard_bounds.setdefault(varname.replace('-mean-bins', ''), None)
if '_gene' in varname:
no_labels = True
if 'j_' not in varname:
cwidth, cheight = 1000, 500
line_width_override = 1
elif 'mute-freqs/v' in args.plotdirs[0] or 'mute-freqs/j' in args.plotdirs[0]:
cwidth, cheight = 1000, 500
bounds = plotconfig.default_hard_bounds.setdefault(utils.unsanitize_name(varname.replace('-mean-bins', '')), None)
if 'IGH' in varname:
if 'mute-freqs' in args.plotdirs[0]:
gene = utils.unsanitize_name(simplevarname)
plottitle = gene # + ' -- mutation frequency'
xtitle = 'position'
if utils.get_region(gene) == 'j':
translegend = (0.1, 0.) #(-0.35, -0.02)
else:
translegend = (0.15, -0.02)
xline = None
if utils.get_region(gene) == 'v' and args.cyst_positions is not None:
xline = args.cyst_positions[gene]['cysteine-position']
# normalization_bounds = (int(cyst_positions[gene]['cysteine-position']) - 70, None)
elif utils.get_region(gene) == 'j' and args.tryp_positions is not None:
xline = int(args.tryp_positions[gene])
# normalization_bounds = (None, int(tryp_positions[gene]) + 5)
else:
ilastdash = simplevarname.rfind('-')
gene = utils.unsanitize_name(simplevarname[:ilastdash])
base_varname = simplevarname[ilastdash + 1 :]
base_plottitle = plotconfig.plot_titles[base_varname] if base_varname in plotconfig.plot_titles else ''
plottitle = gene + ' -- ' + base_plottitle
# draw that little #$*(!
linewidths = [line_width_override, ] if line_width_override is not None else args.linewidths
assert args.leaves_per_tree is None
# scale_errors = math.sqrt(args.leaves_per_tree[idir]) if args.leaves_per_tree is not None else args.scale_errors
draw(all_hists[0], var_type, plotname=varname, plotdir=args.outdir, more_hists=all_hists[1:], write_csv=False, stats=args.stats + ' ' + extrastats, bounds=bounds,
shift_overflows=False, errors=(not args.no_errors), scale_errors=args.scale_errors, rebin=args.rebin, plottitle=plottitle, colors=args.colors, linestyles=args.linestyles,
xtitle=xtitle, ytitle=ytitle, xline=xline, draw_str=draw_str, normalize=args.normalize, normalization_bounds=normalization_bounds,
linewidths=linewidths, markersizes=args.markersizes, cwidth=cwidth, cheight=cheight, no_labels=no_labels, graphify=args.graphify, log=log, translegend=translegend)
if not args.dont_calculate_mean_info:
# write mean info
with opener('w')(args.outdir + '/plots/means.csv') as meanfile:
writer = csv.DictWriter(meanfile, ('name', 'means', 'sems', 'normalized-means'))
writer.writeheader()
for ivar in range(len(all_means)):
writer.writerow({
'name':all_names[ivar],
'means':':'.join([str(m) for m in all_means[ivar]]),
'sems':':'.join([str(s) for s in all_sems[ivar]]),
'normalized-means':':'.join([str(nm) for nm in all_normalized_means[ivar]])
})
check_call(['./bin/permissify-www', args.outdir]) # NOTE this should really permissify starting a few directories higher up
check_call(['./bin/makeHtml', args.outdir, '3', 'null', 'svg'])