def plot_hypothesis(hypothesis, file_name):
bin_types = list(hypothesis)
scores = list(hypothesis[bin_types[0]])
plots = []
for bin_type, score in product(bin_types, scores):
mean_name = "Mean: " + score
df = pd.DataFrame(columns=["Bin", "Dataset", mean_name])
df2 = pd.DataFrame(columns=["Bin", "t-statistic", 'p-value'])
for bin_ in hypothesis[bin_type][score]:
h = list(bin_.values())[0]
bin_name = list(bin_)[0]
parameter1 = h.p1
parameter2 = h.p2
mean1 = h.mean1
mean2 = h.mean2
row1 = {
"Bin": bin_name,
'Dataset': parameter1,
mean_name: str(round(float(mean1), 3))
}
row2 = {
"Bin": bin_name,
'Dataset': parameter2,
mean_name: str(round(float(mean2), 3))
}
df = df.append(row1, ignore_index=True)
df = df.append(row2, ignore_index=True)
t_statistic = h.t
p_value = h.p
row = {
"Bin":
bin_name,
't-statistic':
str(round(t_statistic, 3)),
'p-value':
str(p_value),
'95% Confidence':
"Significant" if p_value <= 0.05 else "Not Significant"
}
df2 = df2.append(row, ignore_index=True)
plots.append(
(ggplot(df, aes(x='Bin', y=mean_name, fill='Dataset')) +
geom_col(stat='identity', position='dodge') +
ggtitle("{0} bin distribution| {1}\nBin's Average Scores".format(
bin_type, score))))
plots.append(
(ggplot(df2, aes(x='Bin', y='p-value', fill='95% Confidence')) +
geom_col(stat='identity', width=0.2) + ggtitle(
"{0} bin distribution| {1}\nBin's 95% Confidence Level Test".
format(bin_type, score)) +
scale_fill_manual(values={
'Significant': "#214517",
'Not Significant': '#c62f2d'
})))
save_as_pdf_pages(plots, file_name)
return
def plots_by_site(self, as_pdf=True, filename="plot/figs/All_sites_by_plot.pdf"):
self.opts["plt_group_by"] = ["site", "plot", "julian", "type"]
if self.plot_data is None:
self.create_plot_data()
sites = self.plot_data["site"].unique()
# Update plot options
plot_options = self.get_plot_options()
plot_options["colour"] = "plot"
plot_options["facet_by"] = "plot"
plots = [self.plot_plots(site, plot_options, as_pdf) for site in sites]
plots = [plot for plot in plots if plot is not None]
if as_pdf:
save_as_pdf_pages(plots, filename)
def save_as_pdf(plot: p9.ggplot,
filename: str = None,
path: str = None,
dpi: int = None,
verbose: bool = False) -> None:
"""Save a plotnine ggplot as pdf
Parameters
----------
plot : p9.ggplot
The plot to save
filename : str, optional (default: None)
Filename to write to. If None, a name is generated.
path : str, optional (default: None)
Path to save to. If None, saves to "out".
dpi : int, optional (default: None)
DPI of saved plot. If None, set to 300.
verbose : bool, optional (default: False)
Print progress report.
Returns
-------
None
"""
if path is None:
path = "out"
if filename is None:
dateTimeObj = datetime.now()
filename = "{}/cellex_plot_{}.pdf".format(
path, dateTimeObj.strftime("%y%m%d_%H%M%S"))
if dpi is None:
dpi = 300
os.makedirs(path, exist_ok=True) # make dir if it doesn't already exist
p9.save_as_pdf_pages(plots=[plot], filename=(filename), dpi=dpi)
if verbose:
print("Saved: {}".format(filename))
def ologram_merge_stats(inputfiles=None,
pdf_width=None,
pdf_height=None,
output=None,
labels=None):
# -------------------------------------------------------------------------
# Check user provided labels
# -------------------------------------------------------------------------
if labels is not None:
labels = labels.split(",")
for elmt in labels:
if not re.search("^[A-Za-z0-9_]+$", elmt):
message(
"Only alphanumeric characters and '_' allowed for --more-bed-labels",
type="ERROR")
if len(labels) != len(inputfiles):
message("--labels: the number of labels should be"
" the same as the number of input files ", type="ERROR")
if len(labels) != len(set(labels)):
message("Redundant labels not allowed.", type="ERROR")
# -------------------------------------------------------------------------
# Loop over input files
# -------------------------------------------------------------------------
df_list = list()
df_label = list()
for pos, infile in enumerate(inputfiles):
message("Reading file : " + infile.name)
# Read the dataset into a temporay dataframe
df_tmp = pd.read_csv(infile, sep='\t', header=0, index_col=None)
# Change name of 'feature_type' column.
df_tmp = df_tmp.rename(index=str, columns={"feature_type": "Feature"})
# Assign the name of the dataset to a new column
if labels is None:
file_short_name = os.path.basename(os.path.normpath(os.path.dirname(infile.name)))
df_label += [file_short_name]
else:
file_short_name = labels[pos]
df_label += [labels[pos]]
df_tmp = df_tmp.assign(**{"dataset": [file_short_name] * df_tmp.shape[0]})
# Pval set to 0 or -1 are changed to 1e-320 and NaN respectively
df_tmp.loc[df_tmp['summed_bp_overlaps_pvalue'] == 0, 'summed_bp_overlaps_pvalue'] = 1e-320
df_tmp.loc[df_tmp['summed_bp_overlaps_pvalue'] == -1, 'summed_bp_overlaps_pvalue'] = np.nan
# Compute -log10(pval)
df_tmp = df_tmp.assign(**{"-log_10(pval)": -np.log10(df_tmp.summed_bp_overlaps_pvalue)})
# Which p-values are signifcant ?
# TODO: For now, draws all p-values. Add Benjamini-Hochberg correction, and distinguish between NaN and 0.
df_tmp = df_tmp.assign(**{"pval_signif": df_tmp.summed_bp_overlaps_pvalue > 0})
# Add the df to the list to be subsequently merged
df_list += [df_tmp]
if len(set(df_label)) < len(df_label):
message('Enclosing directories are ambiguous and cannot be used as labels. You may use "--labels".',
type="ERROR")
# -------------------------------------------------------------------------
# Concatenate dataframes (row bind)
# -------------------------------------------------------------------------
message("Merging dataframes.")
df_merged = pd.concat(df_list, axis=0)
# -------------------------------------------------------------------------
# Plotting
# -------------------------------------------------------------------------
message("Plotting")
my_plot = ggplot(data=df_merged,
mapping=aes(y='Feature', x='dataset'))
my_plot += geom_tile(aes(fill = 'summed_bp_overlaps_log2_fold_change'))
my_plot += scale_fill_gradient2()
my_plot += labs(fill = "log2(fold change) for summed bp overlaps")
# Points for p-val. Must be after geom_tile()
my_plot += geom_point(data = df_merged.loc[df_merged['pval_signif']],
mapping = aes(x='dataset',y='Feature',color = '-log_10(pval)'), size=4, shape ='D', inherit_aes = False)
my_plot += scale_color_gradientn(colors = ["#160E00","#FFB025","#FFE7BD"])
my_plot += labs(color = "-log10(p-value)")
# Theming
my_plot += theme_bw()
my_plot += theme(panel_grid_major=element_blank(),
axis_text_x=element_text(rotation=90),
panel_border=element_blank(),
axis_ticks=element_blank())
# -------------------------------------------------------------------------
# Saving
# -------------------------------------------------------------------------
message("Saving")
nb_ft = len(list(df_merged['Feature'].unique()))
nb_datasets = len(list(df_merged['dataset'].unique()))
if pdf_width is None:
panel_width = 0.6
pdf_width = panel_width * nb_datasets
if pdf_width > 100:
pdf_width = 100
message("Setting --pdf-width to 100 (limit)")
if pdf_height is None:
panel_height = 0.6
pdf_height = panel_height * nb_ft
if pdf_height > 500:
pdf_height = 500
message("Setting --pdf-height to 500 (limit)")
message("Page width set to " + str(pdf_width))
message("Page height set to " + str(pdf_height))
figsize = (pdf_width, pdf_height)
# -------------------------------------------------------------------------
# Turn warning off. Both pandas and plotnine use warnings for deprecated
# functions. I need to turn they off although I'm not really satisfied with
# this solution...
# -------------------------------------------------------------------------
def fxn():
warnings.warn("deprecated", DeprecationWarning)
# -------------------------------------------------------------------------
# Saving
# -------------------------------------------------------------------------
with warnings.catch_warnings():
warnings.simplefilter("ignore")
fxn()
message("Saving diagram to file : " + output.name)
message("Be patient. This may be long for large datasets.")
# NOTE : We must manually specify figure size with save_as_pdf_pages
save_as_pdf_pages(filename=output.name,
plots=[my_plot + theme(figure_size=figsize)],
width=pdf_width,
height=pdf_height)
}).reset_index().astype({"tag": "category"}))
plt = ggplot(
data=tags_summary,
mapping=aes(
x="tag",
y="n_tags",
), # "factor(species, ordered=False)",
)
plt = (
plt + geom_bar(
stat="identity", show_legend=True, position=position_dodge()) +
xlab("Species") + ylab("Number of annotations") +
geom_text(mapping=aes(label="n_tags"),
position=position_dodge(width=0.9)) + theme_classic() +
theme(
axis_text_x=element_text(
angle=90, vjust=1, hjust=1, margin={"r": -30}),
figure_size=(20, 8),
) +
ggtitle("_".join([database["name"], db_type, "tag_species.png"]) +
"(n = " + str(tag_df.shape[0]) + ")")
# + scale_x_discrete(limits=SPECIES_LIST, labels=xlabels)
)
plots.append(plt)
# plt.save(
# ), width=10, height=8
# )
# print(tags_summary)
save_as_pdf_pages(plots, "tag_summaries3_" + opts["class_type"] + ".pdf")
+p9.geom_abline(p9.aes(intercept=wyn['Intercept'],slope=wyn["btystdave"])))
print(fig1)
figures.append(fig1)
df2=beauty
df2['y_pred']=results.predict()
df2['residuals']=df2['courseevaluation']-df2['y_pred']
fig2_res=(p9.ggplot(p9.aes(x='btystdave',y='residuals'),data=beauty)
+p9.geom_point())
print(fig2_res)
figures.append(fig2_res)
results = smf.ols("courseevaluation" +"~btystdavepos + btystdave", data=dane).fit()
wyn=results.params
fig2=(p9.ggplot(p9.aes(x="btystdave",y="courseevaluation"),data=dane)
+p9.geom_jitter(width=0.1)
+p9.geom_abline(p9.aes(intercept=wyn['Intercept'],slope=wyn["btystdave"])))
print(fig2)
figures.append(fig2)
df2=beauty
df2['y_pred']=results.predict()
df2['residuals']=df2['courseevaluation']-df2['y_pred']
fig2_res=(p9.ggplot(p9.aes(x='btystdave',y='residuals'),data=beauty)
+p9.geom_point())
print(fig2_res)
figures.append(fig2_res)
save_as_pdf_pages(figures, filename="./zad2results/figures.pdf")
max_avg_dist_df, aes(x="Metric", y="Distance", fill="Dataset")
) + geom_bar(stat='identity', position='dodge') + labs(
title=
"Feature Selection {} | Maximum Classifier Score | Average Error Distance"
.format(fs_method), ) + ylab("Average Error")))
plots.append((ggplot(
max_max_dist_df, aes(x="Metric", y="Distance", fill="Dataset")
) + geom_bar(stat='identity', position='dodge') + labs(
title=
"Feature Selection {} | Maximum Classifier Score | Maximum Error Distance"
.format(fs_method), ) + ylab("Max Error")))
plots.append((ggplot(
mean_avg_dist_df, aes(x="Metric", y="Distance", fill="Dataset")
) + geom_bar(stat='identity', position='dodge') + labs(
title=
"Feature Selection {} | Average Classifier Score | Average Error Distance"
.format(fs_method), ) + ylab("Avg Error")))
plots.append((ggplot(
mean_max_dist_df, aes(x="Metric", y="Distance", fill="Dataset")
) + geom_bar(stat='identity', position='dodge') + labs(
title=
"Feature Selection {} | Average Classifier Score | Maximum Error Distance"
.format(fs_method), ) + ylab("Max Error")))
path = Config.get_work_dir_path(
os.path.join("paper", "hypothesis", "robustness.pdf"))
save_as_pdf_pages(plots, path=path)
+ gg.ylab("UMAP Y")
+ gg.theme(
legend_position="none",
strip_text=gg.element_text(size=5),
strip_background=gg.element_rect(colour="black", fill="#fdfff4"),
axis_text=gg.element_text(size=6),
axis_title=gg.element_text(size=7),
title=gg.element_text(size=7),
figure_size=(5.5, 3)
)
)
plotlist.append(embedding_gg)
output_file = pathlib.Path(output_dir[batch], f"{batch}_UMAPs.pdf")
output_file.parent.mkdir(exist_ok=True)
gg.save_as_pdf_pages(plotlist, output_file)
# In[6]:
batch = "2017_12_05_Batch2"
plotlist = []
for norm_method in norm_methods:
for color_type in [
"Metadata_broad_sample", "Metadata_Plate", "Metadata_cell_line", "Metadata_time_point"
]:
output_file = pathlib.Path(output_dir[batch], f"{batch}_{norm_method}_colorby{color_type}.png")
output_file.parent.mkdir(exist_ok=True)
fig2 = (p9.ggplot(p9.aes(x='x2', y='y'), data=excersise) +
p9.geom_jitter(width=0.1) +
p9.geom_abline(p9.aes(intercept=wyn['Intercept'], slope=wyn['x2'])))
plots.append(fig2)
print(fig2)
df = excersise
df['excersise_predict'] = results.predict()
df['residuals'] = df['y'] - df['excersise_predict']
fig1_res = (p9.ggplot(p9.aes(x='x1', y='residuals'), data=excersise) +
p9.geom_point())
print(fig1_res)
plots.append(fig1_res)
fig2_res = (p9.ggplot(p9.aes(x='x2', y='residuals'), data=excersise) +
p9.geom_point())
print(fig2_res)
plots.append(fig2_res)
save_as_pdf_pages(plots, filename="./zad1results/figures.pdf")
exercise_v2 = pd.read_csv("exercise.csv")
results_v2 = smf.ols('y ~ x1 + x2', data=exercise_v2).fit()
X = exercise_v2[40:60].drop('y', axis=1)
Y_pred = results_v2.predict(X)
plt.plot(Y_pred)
plt.savefig('./zad1results/prediction.png')
plt.show()