def getStats(self):
'''
Calculates all different kind of stats for each taxonomic group based on each BLAST file and the consensus...
'''
self.stats['count'] = {}
self.stats['span'] = {}
self.stats['n50'] = {}
self.stats['lengths'] = {}
self.stats['gc'] = {}
self.stats['cov'] = {}
self.stats['total_count'] = 0
self.stats['total_span'] = 0
self.stats['total_n50'] = 0
self.stats['total_lengths'] = []
self.stats['total_cov'] = {}
self.stats['total_gc'] = {'raw': [], 'mean': 0.0, 'stdev': 0.0}
self.stats['cov_libs'] = []
for contig_name in self.contigs:
blob = self.contigs[contig_name]
self.stats['total_count'] += 1
self.stats['total_span'] += blob.length
self.stats['total_lengths'].append(blob.length)
self.stats['total_gc']['raw'].append(blob.gc)
for blast_lib in self.blast_libs:
bestTax = keyWithMaxVal(blob.tax[blast_lib])
if not blast_lib in self.stats['count']:
self.stats['count'][blast_lib] = {}
self.stats['span'][blast_lib] = {}
self.stats['lengths'][blast_lib] = {}
self.stats['gc'][blast_lib] = {}
self.stats['cov'][blast_lib] = {}
self.stats['count'][
blast_lib][bestTax] = self.stats['count'][blast_lib].get(
bestTax, 0) + 1
self.stats['span'][
blast_lib][bestTax] = self.stats['span'][blast_lib].get(
bestTax, 0) + blob.length
if not bestTax in self.stats['gc'][blast_lib]:
self.stats['gc'][blast_lib][bestTax] = {
'raw': [],
'mean': 0.0,
'stdev': 0.0
}
self.stats['lengths'][blast_lib][bestTax] = []
self.stats['gc'][blast_lib][bestTax]['raw'].append(blob.gc)
self.stats['lengths'][blast_lib][bestTax].append(blob.length)
for cov_lib, cov in blob.covs.items():
if not cov_lib in self.stats['cov'][blast_lib]:
self.stats['total_cov'][cov_lib] = {
'raw': [],
'mean': 0.0,
'stdev': 0.0
}
self.stats['cov'][blast_lib][cov_lib] = {}
if not bestTax in self.stats['cov'][blast_lib][cov_lib]:
self.stats['cov'][blast_lib][cov_lib][bestTax] = {
'raw': [],
'mean': 0.0,
'stdev': 0.0
}
self.stats['cov'][blast_lib][cov_lib][bestTax][
'raw'].append(cov)
for cov_lib, cov in blob.covs.items():
self.stats['total_cov'][cov_lib]['raw'].append(cov)
for blast_lib in self.blast_libs:
# calculate N50
for tax, list_of_lengths in self.stats['lengths'][blast_lib].items(
):
if not blast_lib in self.stats['n50']:
self.stats['n50'][blast_lib] = {}
self.stats['n50'][blast_lib][tax] = n50(list_of_lengths)
self.stats['total_n50'] = n50(self.stats['total_lengths'])
# calculate total gc mean/stdev
for tax in self.stats['gc'][blast_lib]:
self.stats['gc'][blast_lib][tax]['mean'] = "{0:.2f}".format(
numpy.mean(self.stats['gc'][blast_lib][tax]['raw']))
self.stats['gc'][blast_lib][tax]['stdev'] = "{0:.2f}".format(
numpy.std(self.stats['gc'][blast_lib][tax]['raw']))
# calculate total cov mean/stdev
for cov_lib in self.stats['cov'][blast_lib]:
self.stats['total_cov'][cov_lib]['mean'] = "{0:.2f}".format(
numpy.mean(self.stats['total_cov'][cov_lib]['raw']))
self.stats['total_cov'][cov_lib]['stdev'] = "{0:.2f}".format(
numpy.std(self.stats['total_cov'][cov_lib]['raw']))
# calculate tax-specific cov mean/stdev
for tax in self.stats['cov'][blast_lib][cov_lib]:
self.stats['cov'][blast_lib][cov_lib][tax][
'mean'] = "{0:.2f}".format(
numpy.mean(self.stats['cov'][blast_lib][cov_lib]
[tax]['raw']))
self.stats['cov'][blast_lib][cov_lib][tax][
'stdev'] = "{0:.2f}".format(
numpy.std(self.stats['cov'][blast_lib][cov_lib]
[tax]['raw']))
self.stats['total_gc']['mean'] = "{0:.2f}".format(
numpy.mean(self.stats['total_gc']['raw']))
self.stats['total_gc']['stdev'] = "{0:.2f}".format(
numpy.std(self.stats['total_gc']['raw']))
def plot(data, cov_data, outfile, title):
""" Plotting function which gets masked data and plots to outfile"""
rect_scatter, rect_histx, rect_histy, rect_legend = set_canvas()
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
# Setting up plots and axes
plt.figure(1, figsize=(35,35), dpi=400)
axScatter = plt.axes(rect_scatter, axisbg=background_grey, yscale = 'log')
axScatter = set_format_scatterplot(axScatter)
axHistx = plt.axes(rect_histx, axisbg=background_grey)
axHistx = set_format_hist_x(axHistx, axScatter)
axHisty = plt.axes(rect_histy, axisbg=background_grey)
axHisty = set_format_hist_y(axHisty, axScatter)
axScatter.yaxis.get_major_ticks()[0].label1.set_visible(False)
axScatter.yaxis.get_major_ticks()[1].label1.set_visible(False)
if (title):
plt.suptitle(title, fontsize=35, verticalalignment='top')
#plt.suptitle(out_file, fontsize=25, verticalalignment='bottom')
axLegend = plt.axes(rect_legend, axisbg=white)
axLegend.xaxis.set_major_locator(plt.NullLocator())
axLegend.xaxis.set_major_formatter(nullfmt)
axLegend.yaxis.set_major_locator(plt.NullLocator())
axLegend.yaxis.set_major_formatter(nullfmt)
#
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
# Setting bins for histograms
top_bins = np.arange(0, 1, 0.01)
right_bins = np.logspace(-2, (int(math.log(max_cov)) + 1), 200, base=10.0)
# empty handles for big legend
legend_handles = []
legend_labels = []
# change file name if span (should be in input parsing function)
if hist_span:
outfile += ".hist_span"
else:
outfile += ".hist_count"
# counter necessary for multiplot so that PNGs are in order when sorted by name
i = 0
# initiate variables for plotting
s, lw, alpha, color = 0, 0, 0, ''
# Maybe make an STDOUT printing func?
print "[STATUS] Plotting : " + outfile
# for each phylum ... they are ordered
for tax in tax_list:
i += 1
# get indices for those rows in data where the phylum == tax
index_for_tax = np.where(data[:,3].astype(str) == tax, True, False)
# count of contigs ... total number of contigs comes from previous step?
number_of_contigs_for_tax = np.sum(index_for_tax)
# uses number_of_contigs for checking whether plotting should be carried out ... maybe there is a better place for this ...
if number_of_contigs_for_tax == 0:
pass
else:
# sums span for phylum in mb and not ... do we need both?
span_of_contigs_for_tax = np.sum(data[index_for_tax][:,1].astype(int))
span_of_contigs_for_tax_in_mb = span_of_contigs_for_tax/1000000
# create np_arrays for length, gc and cov for all contigs in phylum
len_array = data[index_for_tax][:,1].astype(int)
gc_array = data[index_for_tax][:,2].astype(float)
cov_array = cov_data[index_for_tax].astype(float)
# generates label ... this should be turned into a table ...
label = tax + " (" + "{:,}".format(number_of_contigs_for_tax) + "; " + "%.2f" % round(span_of_contigs_for_tax_in_mb,2) + "MB; " + "{:,}".format(n50(len_array)) + "nt)"
# another status message
print "\t" + label
s_array = []
# ignore contig length ... maybe do this in input and set these params for plotting there ...
if (ignore_contig_len):
if tax == 'no-hit':
s, lw, alpha, color = 15, 0.5, 0.5, grey
else:
s, lw, alpha, color = 65, 0.5, 1, color_dict[tax]
s_array = [s for contig_length in len_array]
else:
if tax == 'no-hit':
s, lw, alpha, color = 15, 0.5, 0.5, grey
else:
s, lw, alpha, color = 15, 0.5, 1, color_dict[tax]
# these are the sizes for plotting with contig sizes
s_array = [contig_length/s for contig_length in len_array]
# making copies of gc/cov_array
gc_hist_array = gc_array
cov_hist_array = cov_array
#######
# if hist span ...
# make a new array ...
# add to the array : (gc * len/1000) - 1
# substitute old array with new array
# set histogram labels depending on type ... can be set before ...
weights_array = len_array/1000
if (hist_span):
axHistx.set_ylabel("Span (kb)")
axHisty.set_xlabel("Span (kb)", rotation='horizontal')
else:
axHistx.set_ylabel("Count")
axHisty.set_xlabel("Count", rotation='horizontal')
# this should be set before ... or after? ... but only once
for xtick in axHisty.get_xticklabels(): # rotate text for ticks in cov histogram
xtick.set_rotation(270)
# add text to legend ... label was build before ... could be a function
legend_handles.append(Line2D([0], [0], linewidth = 0.5, linestyle="none", marker="o", alpha=1, markersize=24, markerfacecolor=color))
legend_labels.append(label)
if (number_of_contigs_for_tax):
if (hist_span):
axHistx.hist(gc_hist_array, weights=weights_array , color = color, bins = top_bins, histtype='step', lw = 3)
axHisty.hist(cov_hist_array, weights=weights_array , color = color, bins = right_bins, histtype='step', orientation='horizontal', lw = 3)
else:
axHistx.hist(gc_hist_array, color = color, bins = top_bins, histtype='step', lw = 3)
axHisty.hist(cov_hist_array , color = color, bins = right_bins, histtype='step', orientation='horizontal', lw = 3)
axScatter.scatter(gc_array, cov_array, color = color, s = s_array, lw = lw, alpha=alpha, edgecolor=black, label=label)
axLegend.axis('off')
if (multi_plot): # MULTI-PLOT!!!
axLegend.legend(legend_handles, legend_labels, loc=6, numpoints=1, fontsize=fontsize, frameon=True)
plot_ref_legend(axScatter)
#plt.savefig(outfile + "." + str(i) + "_"+ tax.replace("/","") + "." + fig_format, format=fig_format)
plt.savefig(outfile + "." + str(i) + "_"+ tax.replace("/","") + "." + fig_format, format=fig_format)
if ignore_contig_len:
pass
else: # print scale-legend
plot_ref_legend(axScatter)
axLegend.legend(legend_handles, legend_labels, numpoints=1, fontsize=fontsize, frameon=True, loc=6 )
sys.stdout.write("Saving file " + outfile)
plt.savefig(outfile + "." + fig_format, format=fig_format)
plt.close()
print " [Done]\n"
def getStats(self):
'''
Calculates all different kind of stats for each taxonomic group based on each BLAST file and the consensus...
'''
self.stats['count'] = {}
self.stats['span']= {}
self.stats['n50']= {}
self.stats['lengths']= {}
self.stats['gc']= {}
self.stats['cov'] = {}
self.stats['total_count'] = 0
self.stats['total_span'] = 0
self.stats['total_n50'] = 0
self.stats['total_lengths'] = []
self.stats['total_cov'] = {}
self.stats['total_gc'] = {'raw' : [], 'mean' : 0.0, 'stdev' : 0.0}
self.stats['cov_libs'] = []
for contig_name in self.contigs:
blob = self.contigs[contig_name]
self.stats['total_count'] += 1
self.stats['total_span'] += blob.length
self.stats['total_lengths'].append(blob.length)
self.stats['total_gc']['raw'].append(blob.gc)
for blast_lib in self.blast_libs:
bestTax = keyWithMaxVal(blob.tax[blast_lib])
if not blast_lib in self.stats['count']:
self.stats['count'][blast_lib] = {}
self.stats['span'][blast_lib] = {}
self.stats['lengths'][blast_lib] = {}
self.stats['gc'][blast_lib] = {}
self.stats['cov'][blast_lib] = {}
self.stats['count'][blast_lib][bestTax] = self.stats['count'][blast_lib].get(bestTax, 0) + 1
self.stats['span'][blast_lib][bestTax] = self.stats['span'][blast_lib].get(bestTax, 0) + blob.length
if not bestTax in self.stats['gc'][blast_lib]:
self.stats['gc'][blast_lib][bestTax] = {'raw' : [], 'mean' : 0.0, 'stdev' : 0.0}
self.stats['lengths'][blast_lib][bestTax] = []
self.stats['gc'][blast_lib][bestTax]['raw'].append(blob.gc)
self.stats['lengths'][blast_lib][bestTax].append(blob.length)
for cov_lib, cov in blob.covs.items():
if not cov_lib in self.stats['cov'][blast_lib]:
self.stats['total_cov'][cov_lib] = {'raw' : [], 'mean' : 0.0, 'stdev' : 0.0}
self.stats['cov'][blast_lib][cov_lib]={}
if not bestTax in self.stats['cov'][blast_lib][cov_lib]:
self.stats['cov'][blast_lib][cov_lib][bestTax] = {'raw' : [], 'mean' : 0.0, 'stdev' : 0.0}
self.stats['cov'][blast_lib][cov_lib][bestTax]['raw'].append(cov)
for cov_lib, cov in blob.covs.items():
self.stats['total_cov'][cov_lib]['raw'].append(cov)
for blast_lib in self.blast_libs:
# calculate N50
for tax, list_of_lengths in self.stats['lengths'][blast_lib].items():
if not blast_lib in self.stats['n50']:
self.stats['n50'][blast_lib] = {}
self.stats['n50'][blast_lib][tax] = n50(list_of_lengths)
self.stats['total_n50'] = n50(self.stats['total_lengths'])
# calculate total gc mean/stdev
for tax in self.stats['gc'][blast_lib]:
self.stats['gc'][blast_lib][tax]['mean'] = "{0:.2f}".format(numpy.mean(self.stats['gc'][blast_lib][tax]['raw']))
self.stats['gc'][blast_lib][tax]['stdev'] = "{0:.2f}".format(numpy.std(self.stats['gc'][blast_lib][tax]['raw']))
# calculate total cov mean/stdev
for cov_lib in self.stats['cov'][blast_lib]:
self.stats['total_cov'][cov_lib]['mean'] = "{0:.2f}".format(numpy.mean(self.stats['total_cov'][cov_lib]['raw']))
self.stats['total_cov'][cov_lib]['stdev'] = "{0:.2f}".format(numpy.std(self.stats['total_cov'][cov_lib]['raw']))
# calculate tax-specific cov mean/stdev
for tax in self.stats['cov'][blast_lib][cov_lib]:
self.stats['cov'][blast_lib][cov_lib][tax]['mean'] = "{0:.2f}".format(numpy.mean(self.stats['cov'][blast_lib][cov_lib][tax]['raw']))
self.stats['cov'][blast_lib][cov_lib][tax]['stdev'] = "{0:.2f}".format(numpy.std(self.stats['cov'][blast_lib][cov_lib][tax]['raw']))
self.stats['total_gc']['mean'] = "{0:.2f}".format(numpy.mean(self.stats['total_gc']['raw']))
self.stats['total_gc']['stdev'] = "{0:.2f}".format(numpy.std(self.stats['total_gc']['raw']))
def plot(data, cov_data, outfile, title):
""" Plotting function which gets masked data and plots to outfile"""
rect_scatter, rect_histx, rect_histy, rect_legend = set_canvas()
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
# Setting up plots and axes
plt.figure(1, figsize=(35, 35), dpi=400)
axScatter = plt.axes(rect_scatter, axisbg=background_grey, yscale='log')
axScatter = set_format_scatterplot(axScatter)
axHistx = plt.axes(rect_histx, axisbg=background_grey)
axHistx = set_format_hist_x(axHistx, axScatter)
axHisty = plt.axes(rect_histy, axisbg=background_grey)
axHisty = set_format_hist_y(axHisty, axScatter)
axScatter.yaxis.get_major_ticks()[0].label1.set_visible(False)
axScatter.yaxis.get_major_ticks()[1].label1.set_visible(False)
if (title):
plt.suptitle(title, fontsize=35, verticalalignment='top')
#plt.suptitle(out_file, fontsize=25, verticalalignment='bottom')
axLegend = plt.axes(rect_legend, axisbg=white)
axLegend.xaxis.set_major_locator(plt.NullLocator())
axLegend.xaxis.set_major_formatter(nullfmt)
axLegend.yaxis.set_major_locator(plt.NullLocator())
axLegend.yaxis.set_major_formatter(nullfmt)
#
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
# Setting bins for histograms
top_bins = np.arange(0, 1, 0.01)
right_bins = np.logspace(-2, (int(math.log(max_cov)) + 1), 200, base=10.0)
# empty handles for big legend
legend_handles = []
legend_labels = []
# change file name if span (should be in input parsing function)
if hist_span:
outfile += ".hist_span"
else:
outfile += ".hist_count"
# counter necessary for multiplot so that PNGs are in order when sorted by name
i = 0
# initiate variables for plotting
s, lw, alpha, color = 0, 0, 0, ''
# Maybe make an STDOUT printing func?
print "[STATUS] Plotting : " + outfile
# for each phylum ... they are ordered
for tax in tax_list:
i += 1
# get indices for those rows in data where the phylum == tax
index_for_tax = np.where(data[:, 3].astype(str) == tax, True, False)
# count of contigs ... total number of contigs comes from previous step?
number_of_contigs_for_tax = np.sum(index_for_tax)
# uses number_of_contigs for checking whether plotting should be carried out ... maybe there is a better place for this ...
if number_of_contigs_for_tax == 0:
pass
else:
# sums span for phylum in mb and not ... do we need both?
span_of_contigs_for_tax = np.sum(
data[index_for_tax][:, 1].astype(int))
span_of_contigs_for_tax_in_mb = span_of_contigs_for_tax / 1000000
# create np_arrays for length, gc and cov for all contigs in phylum
len_array = data[index_for_tax][:, 1].astype(int)
gc_array = data[index_for_tax][:, 2].astype(float)
cov_array = cov_data[index_for_tax].astype(float)
# generates label ... this should be turned into a table ...
label = tax + " (" + "{:,}".format(
number_of_contigs_for_tax) + "; " + "%.2f" % round(
span_of_contigs_for_tax_in_mb, 2) + "MB; " + "{:,}".format(
n50(len_array)) + "nt)"
# another status message
print "\t" + label
s_array = []
# ignore contig length ... maybe do this in input and set these params for plotting there ...
if (ignore_contig_len):
if tax == 'no-hit':
s, lw, alpha, color = 15, 0.5, 0.5, grey
else:
s, lw, alpha, color = 65, 0.5, 1, color_dict[tax]
s_array = [s for contig_length in len_array]
else:
if tax == 'no-hit':
s, lw, alpha, color = 15, 0.5, 0.5, grey
else:
s, lw, alpha, color = 15, 0.5, 1, color_dict[tax]
# these are the sizes for plotting with contig sizes
s_array = [contig_length / s for contig_length in len_array]
# making copies of gc/cov_array
gc_hist_array = gc_array
cov_hist_array = cov_array
#######
# if hist span ...
# make a new array ...
# add to the array : (gc * len/1000) - 1
# substitute old array with new array
# set histogram labels depending on type ... can be set before ...
weights_array = len_array / 1000
if (hist_span):
axHistx.set_ylabel("Span (kb)")
axHisty.set_xlabel("Span (kb)", rotation='horizontal')
else:
axHistx.set_ylabel("Count")
axHisty.set_xlabel("Count", rotation='horizontal')
# this should be set before ... or after? ... but only once
for xtick in axHisty.get_xticklabels(
): # rotate text for ticks in cov histogram
xtick.set_rotation(270)
# add text to legend ... label was build before ... could be a function
legend_handles.append(
Line2D([0], [0],
linewidth=0.5,
linestyle="none",
marker="o",
alpha=1,
markersize=24,
markerfacecolor=color))
legend_labels.append(label)
if (number_of_contigs_for_tax):
if (hist_span):
axHistx.hist(gc_hist_array,
weights=weights_array,
color=color,
bins=top_bins,
histtype='step',
lw=3)
axHisty.hist(cov_hist_array,
weights=weights_array,
color=color,
bins=right_bins,
histtype='step',
orientation='horizontal',
lw=3)
else:
axHistx.hist(gc_hist_array,
color=color,
bins=top_bins,
histtype='step',
lw=3)
axHisty.hist(cov_hist_array,
color=color,
bins=right_bins,
histtype='step',
orientation='horizontal',
lw=3)
axScatter.scatter(gc_array,
cov_array,
color=color,
s=s_array,
lw=lw,
alpha=alpha,
edgecolor=black,
label=label)
axLegend.axis('off')
if (multi_plot): # MULTI-PLOT!!!
axLegend.legend(legend_handles,
legend_labels,
loc=6,
numpoints=1,
fontsize=fontsize,
frameon=True)
plot_ref_legend(axScatter)
#plt.savefig(outfile + "." + str(i) + "_"+ tax.replace("/","") + "." + fig_format, format=fig_format)
plt.savefig(outfile + "." + str(i) + "_" +
tax.replace("/", "") + "." + fig_format,
format=fig_format)
if ignore_contig_len:
pass
else: # print scale-legend
plot_ref_legend(axScatter)
axLegend.legend(legend_handles,
legend_labels,
numpoints=1,
fontsize=fontsize,
frameon=True,
loc=6)
sys.stdout.write("Saving file " + outfile)
plt.savefig(outfile + "." + fig_format, format=fig_format)
plt.close()
print " [Done]\n"