def test_get_mit_pc1(self):
"""returns the correct MI terms when the pseudocount=1"""
mit = get_mit(self.ref_data, pseudocount=1,
check_valid=True).round(decimals=5)
expect_mit = array([[0.02377, 0.02377], [0, -0.02388],
[0.09895, -0.01639], [0.09895, 0.22805]])
numpy.testing.assert_equal(mit, expect_mit)
def test_get_mit_pc0(self):
"""returns the correct MI terms when the pseudocount=0"""
mit = get_mit(self.ref_data, pseudocount=0,
check_valid=True).round(decimals=5)
expect_mit = array([[0.03561, 0.03561], [0.05782, 0],
[0.16781, -0.02109], [0.16781, 0.0]])
numpy.testing.assert_equal(mit, expect_mit)
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")