def get_four_position_fig(four_pos_results, positions, figsize,
group_label=None, group_ref=None, figwidth=None,
xtick_fontsize=14, ytick_fontsize=14):
position_sets = list(combinations(positions, 4))
assert len(position_sets) == 1
rel_entropies = [four_pos_results[position_sets[0]]['rel_entropy']]
ylim = logo.est_ylim(rel_entropies)
rel_entropy = rel_entropies[0]
fig = pyplot.figure(figsize=figsize)
ax = fig.gca()
num_pos = len(positions) + 1
mid_pos = num_pos // 2
position_re = numpy.zeros((num_pos,), float)
characters = numpy.zeros((num_pos, 256), str)
rets = numpy.zeros((256, num_pos), float)
indices = list(range(4))
# now adjust indices to reflect position along sequence
for i in range(len(indices)):
if indices[i] >= mid_pos:
indices[i] += 1
position_re.put(indices, rel_entropy)
stats = four_pos_results[position_sets[0]]['stats']
mut_stats = stats[
get_selected_indices(stats, group_label=group_label,
group_ref=group_ref)][['base1', 'base2', 'base3',
'base4', 'ret']]
mut_stats = mut_stats.sort_values(by='ret')
characters[indices[0]] = list(mut_stats['base1'])
characters[indices[1]] = list(mut_stats['base2'])
characters[indices[2]] = list(mut_stats['base3'])
characters[indices[3]] = list(mut_stats['base4'])
for index in indices:
rets[:, index] = mut_stats['ret']
heights = get_re_char_heights(rets, re_positionwise=position_re)
logo.draw_multi_position(char_heights=heights.T, characters=characters,
position_indices=indices, ax=ax, ylim=ylim,
xtick_fontsize=xtick_fontsize,
ytick_fontsize=ytick_fontsize)
return fig
def draw_position_grid(directions, sample_size=False, width=8, height=8,
title_space=1.1, axis_font_size=20, tick_font_size=10,
ylim=None):
"""docstring for draw_position_grid"""
f, axes = pyplot.subplots(4, 4, sharex=True, sharey=True,
figsize=(width, height))
bases = list('CTAG')
positions = None
plottables = []
adaptive_y = 0
for direction in directions:
data = directions[direction]
if positions is None:
positions = list(data.keys())
positions.sort()
number = data[positions[0]]['stats']["count"].sum() // 2
heights, characters, indices = get_plot_data(data, positions)
adaptive_y = max(adaptive_y, logo.est_ylim(heights))
plottables.append([direction, heights, characters, indices, number])
if ylim is None:
ylim = adaptive_y
for direction, heights, characters, indices, number in plottables:
fr, to = list(map(bases.index, direction.split('to')))
ax = axes[fr, to]
fig = logo.draw_multi_position(heights, characters=characters,
position_indices=indices, ylim=ylim,
ax=ax, figwidth=width, verbose=False)
if sample_size:
y = ax.get_ylim()[1]
ax.text(0.2, y * 0.85, "N={:,}".format(number), fontsize=10)
xformat = FuncFormatter(format_float(1e-3, float_places=2))
for i in range(4):
top_ax = axes[0, i]
top_ax.set_title(
bases[i], fontsize=axis_font_size, weight="bold", y=1.1)
lft_ax = axes[i, 0]
for yticklabel in lft_ax.get_yticklabels():
yticklabel.set_fontsize(tick_font_size)
yticklabel.set_rotation(0)
lft_ax.yaxis.set_major_formatter(FuncFormatter(xformat))
lft_ax.set_ylabel(bases[i], rotation=0,
fontsize=axis_font_size, weight="bold")
lft_ax.yaxis.labelpad = axis_font_size
btm_ax = axes[-1, i]
for xticklabel in btm_ax.get_xticklabels():
xticklabel.set_fontsize(tick_font_size)
xticklabel.set_rotation(0)
f.tight_layout()
return f
def get_single_position_fig(single_results, positions, figsize,
group_label=None, group_ref=None, figwidth=None,
xlabel_fontsize=14, ylabel_fontsize=14,
xtick_fontsize=14, ytick_fontsize=14):
num_pos = len(positions) + 1
mid = num_pos // 2
position_re = numpy.zeros((num_pos,), float)
rets = numpy.zeros((4, num_pos), float)
characters = [list('ACGT') for i in range(num_pos)]
for index, pos in enumerate(positions):
if index >= mid:
index += 1
stats = single_results[pos]['stats']
position_re[index] = single_results[pos]['rel_entropy']
mut_stats = stats[get_selected_indices(stats, group_label=group_label,
group_ref=group_ref)][['base',
'ret']]
mut_stats = mut_stats.sort_values(by='ret')
characters[index] = list(mut_stats['base'])
rets[:, index] = mut_stats['ret']
heights = get_re_char_heights(rets, re_positionwise=position_re)
fig = logo.draw_multi_position(heights.T, characters=characters,
position_indices=list(range(num_pos)),
figsize=figsize, figwidth=figwidth,
verbose=False)
if figwidth:
fig.set_figwidth(figwidth)
ax = fig.gca()
ax.set_xlabel('Position', fontsize=xlabel_fontsize)
ax.set_ylabel('RE', rotation='vertical', fontsize=ylabel_fontsize)
ax.tick_params(axis='x', labelsize=xtick_fontsize, pad=xtick_fontsize // 2,
length=0)
ax.tick_params(axis='y', labelsize=ytick_fontsize, pad=ytick_fontsize // 2)
return fig
def mi(json_path, plot_cfg, no_type3, figpath, format, sample_size,
force_overwrite, dry_run):
"""draws conventional sequence logo, using MI, from first order effects"""
# the following is for logging
json_path = util.abspath(json_path)
args = locals()
if no_type3:
util.exclude_type3_fonts()
if not figpath:
dirname = os.path.dirname(json_path)
figpath = os.path.join(dirname, "MI.%s" % format)
log_file_path = os.path.join(dirname, "MI.log")
else:
figpath = util.abspath(figpath)
log_file_path = "%s.log" % ".".join(figpath.split(".")[:-1])
LOGGER.log_file_path = log_file_path
if plot_cfg:
LOGGER.input_file(plot_cfg)
LOGGER.log_message(str(args), label='vars')
data = util.load_loglin_stats(json_path)
positions = list(data.keys())
positions.sort()
num_pos = len(positions) + 1
mp = num_pos // 2
counts_array = numpy.zeros((4, num_pos), int)
for i, pos in enumerate(positions):
if i >= mp:
i += 1
pos_stats = data[pos]['stats']
counts = pos_stats[pos_stats['mut'] == 'M'][["base", "count"]]
counts = dict(zip(counts['base'], counts['count']))
for base in counts:
base_index = DNA.alphabet.index(base)
counts_array[base_index, i] = counts[base]
freq_matrix = entropy.counts_to_freq_matrix(counts_array)
mit = entropy.get_mit(freq_matrix, freq_matrix=True)
mi = mit.sum(axis=0)
char_hts = get_mi_char_heights(numpy.fabs(mit), mi)
plot_cfg = util.get_plot_configs(cfg_path=plot_cfg)
figsize = plot_cfg.get('1-way plot', 'figsize')
ytick_font = plot_cfg.get('1-way plot', 'ytick_fontsize')
xtick_font = plot_cfg.get('1-way plot', 'xtick_fontsize')
ylabel_font = plot_cfg.get('1-way plot', 'ylabel_fontsize')
xlabel_font = plot_cfg.get('1-way plot', 'xlabel_fontsize')
fig = logo.draw_multi_position(char_hts.T,
characters=[list(DNA)] * num_pos,
position_indices=list(range(num_pos)),
figsize=figsize,
figwidth=figsize[0],
xtick_fontsize=xtick_font,
ytick_fontsize=ytick_font,
sort_data=True)
ax = fig.gca()
ax.tick_params(axis='y', labelsize=ytick_font)
ax.tick_params(axis='x', labelsize=xtick_font)
ax.set_ylabel("MI", fontsize=ylabel_font)
ax.set_xlabel("Position", fontsize=xlabel_font)
fig.tight_layout()
fig.savefig(figpath)
LOGGER.output_file(figpath)
click.secho("Wrote %s" % figpath, fg="green")
def grid(fig_config, figpath, format, no_type3):
"""draws an arbitrary shaped grid of mutation motifs based on fig_config"""
# we read in the config file and determine number of rows and columns
# paths, headings, etc ..
# then create the figure and axes and call the mutation_motif drawing code
args = locals()
if no_type3:
util.exclude_type3_fonts()
if not figpath:
dirname = os.path.dirname(fig_config.name)
figpath = os.path.join(dirname, "drawn_array.%s" % format)
log_file_path = os.path.join(dirname, "drawn_array.log")
else:
figpath = util.abspath(figpath)
log_file_path = "%s.log" % ".".join(figpath.split(".")[:-1])
util.makedirs(os.path.dirname(figpath))
LOGGER.log_file_path = log_file_path
LOGGER.log_message(str(args), label='vars')
ncols, nrows, figsize, col_labels, row_labels, paths, axis_cfg = \
read_plot_array_config(fig_config)
fig, axes = pyplot.subplots(nrows=nrows, ncols=ncols, figsize=figsize,
sharex=True, sharey=True)
figwidth = fig.get_figwidth()
try:
axes[0]
except TypeError:
axes = numpy.array([[axes]])
if len(axes.shape) == 1:
# required for indexing of appropriate axis
axes = numpy.vstack(axes)
if nrows == 1:
axes = axes.T
adaptive_y = 0
plottable = {}
for coord in paths:
data = util.load_loglin_stats(paths[coord])
positions = list(data)
positions.sort()
heights, characters, indices = get_plot_data(data, positions)
adaptive_y = max(adaptive_y, logo.est_ylim(heights))
plottable[coord] = dict(char_heights=heights,
characters=characters,
position_indices=indices,
figwidth=figwidth,
verbose=False)
ylim = axis_cfg.get("ylim", adaptive_y)
for coord in plottable:
kwargs = plottable[coord]
kwargs["ax"] = axes[coord]
kwargs["ylim"] = ylim
fig = logo.draw_multi_position(**kwargs)
xformat = FuncFormatter(format_float(1e-3, float_places=2))
for col in range(ncols):
top_ax = axes[0, col]
top_ax.set_title(col_labels[col], fontsize=axis_cfg["xlabel_fontsize"],
weight="bold", y=1.1)
btm_ax = axes[-1, col]
for xticklabel in btm_ax.get_xticklabels():
xticklabel.set_fontsize(axis_cfg["xtick_fontsize"])
xticklabel.set_rotation(0)
btm_ax.set_xlabel("Position", fontsize=axis_cfg["xlabel_fontsize"],
weight="bold")
btm_ax.xaxis.labelpad = axis_cfg['xlabel_pad']
for row in range(nrows):
lft_ax = axes[row, 0]
for yticklabel in lft_ax.get_yticklabels():
yticklabel.set_fontsize(axis_cfg["ytick_fontsize"])
yticklabel.set_rotation(0)
lft_ax.yaxis.set_major_formatter(FuncFormatter(xformat))
lft_ax.yaxis.labelpad = axis_cfg['ylabel_pad']
lft_ax.set_ylabel(row_labels[row], rotation=0,
fontsize=axis_cfg['ylabel_fontsize'],
weight="bold")
fig.tight_layout()
fig.savefig(figpath)
click.secho("Wrote %s" % figpath, fg="green")
def get_three_position_fig(three_pos_results, positions, figsize,
group_label=None, group_ref=None, figwidth=None,
xtick_fontsize=14, ytick_fontsize=14):
position_sets = list(combinations(positions, 3))
array_coords = get_resized_array_coordinates3(positions, position_sets)
coords = list(array_coords.values())
xdim = max(v[0] for v in coords) + 1
ydim = max(v[1] for v in coords) + 1
fig, axarr = pyplot.subplots(xdim, ydim, figsize=figsize, sharex=True,
sharey=True)
for i in range(xdim):
for j in range(ydim):
if (i, j) in coords:
continue
ax = axarr[i, j]
ax.set_frame_on(False)
ax.get_xaxis().set_visible(False)
ax.get_yaxis().set_visible(False)
num_pos = len(positions) + 1
mid_pos = num_pos // 2
rel_entropies = []
for position_set in position_sets:
rel_entropies.append(three_pos_results[position_set]['rel_entropy'])
ylim = logo.est_ylim(rel_entropies)
position_re = numpy.zeros((num_pos,), float)
multi_positions = {}
characters = numpy.zeros((num_pos, 64), str)
for motif in combinations(positions, 3):
rets = numpy.zeros((64, num_pos), float)
indices = list(map(positions.index, motif))
row, col = array_coords[motif]
ax = axarr[row, col]
# now adjust indices to reflect position along sequence
for i in range(len(indices)):
if indices[i] >= mid_pos:
indices[i] += 1
position_re.put(indices, three_pos_results[motif]['rel_entropy'])
stats = three_pos_results[motif]['stats']
mut_stats = stats[
get_selected_indices(stats, group_label=group_label,
group_ref=group_ref)][['base1', 'base2',
'base3', 'ret']]
mut_stats = mut_stats.sort_values(by='ret')
characters[indices[0]] = list(mut_stats['base1'])
characters[indices[1]] = list(mut_stats['base2'])
characters[indices[2]] = list(mut_stats['base3'])
for index in indices:
rets[:, index] = mut_stats['ret']
heights = get_re_char_heights(rets, re_positionwise=position_re)
multi_positions[motif] = dict(rets=rets, indices=indices,
characters=characters, heights=heights)
logo.draw_multi_position(char_heights=heights.T, characters=characters,
position_indices=indices, ax=ax, ylim=ylim,
xtick_fontsize=xtick_fontsize,
ytick_fontsize=ytick_fontsize)
return fig