def test_re_char_heights(self):
"""char heights for RE terms should be correct"""
rets = array([[0, 0, 0.12535], [0.15782, 0, 0.67927], [0, 0, 0.28571],
[-0.1, 0, -0.21429]])
expect_char_ret_height = array([[0, 0, 0.08417], [0.03539, 0, 0.45612],
[0, 0, 0.19185],
[-0.02243, 0, -0.14389]])
got_ret_height = get_re_char_heights(rets).round(decimals=5)
self.assertEqual(got_ret_height, expect_char_ret_height)
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 get_plot_data(single_results, positions):
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)][['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)
return heights.T, characters, list(range(num_pos))
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 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
