def scrnaseq_results_plot_alignrseqc(df, **kwargs):
"""
Plot scRNASeq QC summary.
Args:
df (DataFrame): data frame consisting of summary data from alignments and RSeQC
kwargs: keyword arguments to bokeh plot functions
Returns:
(dict): dictionary with a bokeh gridplot and a table
"""
df = df.set_index("SM")
source = ColumnDataSource(df)
columns = [
TableColumn(field="SM", title="Sample"),
TableColumn(field="Number_of_input_reads",
title="Number of input reads"),
TableColumn(field="Uniquely_mapped_reads_PCT",
title="Uniquely mapped reads (%)"),
TableColumn(field="Mismatch_rate_per_base,_PCT",
title="Mismatch rate per base (%)"),
TableColumn(field="Insertion_rate_per_base",
title="Insertion rate per base (%)"),
TableColumn(field="Deletion_rate_per_base",
title="Deletion rate per base (%)"),
TableColumn(field="PCT_of_reads_unmapped",
title="Unmapped reads (%)"),
]
table = DataTable(source=source, columns=columns,
editable=False, width=1000)
kwargs = {'plot_width': 400, 'plot_height': 400,
'title_text_font_size': "16pt",
'x_axis_label': 'Sample',
'x_axis_label_text_font_size': '12pt',
'x_major_label_orientation': np.pi/3,
'y_axis_label': 'Reads',
'y_axis_label_text_font_size': '12pt',
'y_major_label_orientation': np.pi/3
}
# Input reads
p1 = dotplot(
df=source,
title="Number of input reads", tools=DEFAULT_TOOLS,
x="SM", y="Number_of_input_reads",
**dict(kwargs, **{'x_range': list(df.index), 'y_axis_type': "log"}))
tooltips(p1, HoverTool, [('Sample', '@SM'),
('Reads', '@Number_of_input_reads')])
# Uniquely mapping
kwargs['y_axis_label'] = 'Percent'
p2 = dotplot(
df=source,
tools=DEFAULT_TOOLS, title="Uniquely mapping reads",
x="SM", y="Uniquely_mapped_reads_PCT",
**dict(kwargs, **{'x_range': p1.x_range, 'y_range': [0, 100]})
)
tooltips(p2, HoverTool, [('Sample', '@SM'),
('Pct_mapped', '@Uniquely_mapped_reads_PCT')])
# Unmapped
p3 = dotplot(
df=source,
tools=DEFAULT_TOOLS, title="Unmapped reads",
x="SM", y="PCT_of_reads_unmapped",
**dict(kwargs, **{'x_range': p1.x_range, 'y_range': [0, 100]})
)
tooltips(p3, HoverTool, [('Sample', '@SM'),
('Pct_unmapped', '@PCT_of_reads_unmapped')])
# Mismatch/indel rate
p4 = dotplot(
df=source,
tools=DEFAULT_TOOLS,
title="Mismatch/indel rate",
x="SM",
y=[
"Mismatch_rate_per_base,_PCT",
"Insertion_rate_per_base",
"Deletion_rate_per_base"
],
**dict(kwargs, **{'x_range': p1.x_range, 'y_range': [0, 1], 'color': ["blue", "red", "green"]})
)
tooltips(p4, HoverTool, [('Sample', '@SM'),
('Mismatch rate per base',
'@Mismatch_rate_per_base,_PCT'),
('Insertion rate per base',
'@Insertion_rate_per_base'),
('Deletion rate per base',
'@Deletion_rate_per_base'), ])
select_tool = p4.select(dict(type=BoxSelectTool))
select_tool.dimensions = ['width']
# Unmapped
p5 = dotplot(
tools=DEFAULT_TOOLS, title="Mismatch/indel sum",
x="SM", y="mismatch_sum", df=source,
**dict(kwargs, **{'x_range': p1.x_range})
)
tooltips(p5, HoverTool, [('Sample', '@SM'),
('Mismatch/indel rate per base',
'@mismatch_sum'), ])
select_tool = p5.select(dict(type=BoxSelectTool))
select_tool.dimensions = ['width']
# Fraction reads mapping to 10% right-most end
p6 = dotplot(
tools=DEFAULT_TOOLS, title="Tags mapping to exons",
x="SM", y="ExonMap_PCT", df=source,
**dict(kwargs, **{'x_range': p1.x_range})
)
tooltips(p6, HoverTool, [('Sample', '@SM'),
('ExonMap (%)', '@ExonMap_PCT'), ])
# Fraction reads mapping to 10% right-most end
p7 = dotplot(
tools=DEFAULT_TOOLS, title="Reads mapping to 3' end",
x="SM", y="three_prime_map", df=source,
**dict(kwargs, **{'x_range': p1.x_range})
)
tooltips(p7, HoverTool, [('Sample', '@SM'),
("3' map", '@three_prime_map'), ])
return {'fig': gridplot([[p1, p2, p3], [p4, p5, p6], [p7]]), 'table': table}
def plot_pca(pcaobjfile, pca_results_file=None, metadata=None, taptool_url=None, **kwargs):
"""Make PCA plot
Args:
pcaobjfile (str): file name containing pickled pca object
pca_results_file (str): pca results file
metadata (str): metadata file name
taptool_url (str): url prefix that is attached to taptool; typically a link to ensembl
Returns:
dict: dictionary with keys 'fig' pointing to a (:py:class:`~bokeh.models.GridPlot`) Bokeh GridPlot object and key 'table' pointing to a (:py:class:`~bokeh.widgets.DataTable`) DataTable
"""
with open(pcaobjfile, 'rb') as fh:
pcaobj = pickle.load(fh)
md = None
if not metadata is None:
md = pd.read_csv(metadata, index_col=0)
df_pca = pd.read_csv(pca_results_file, index_col=kwargs.get('index_col', "SM"))
df_pca['color'] = [kwargs.get('color', 'red')] * df_pca.shape[0]
df_pca['x'] = df_pca['0']
df_pca['y'] = df_pca['1']
df_pca['size'] = [kwargs.get('size', 10)] * df_pca.shape[0]
pca_source = ColumnDataSource(df_pca)
cmap = colorbrewer(datalen = df_pca.shape[0])
callback = CustomJS(args=dict(source=pca_source),
code="""pca_callback(source, cb_obj, "SM");""")
xcallback = CustomJS(args=dict(source=pca_source),
code="""pca_component(source, cb_obj, "x");""")
ycallback = CustomJS(args=dict(source=pca_source),
code="""pca_component(source, cb_obj, "y");""")
if not md is None:
# Waiting for callbacks to be implemented upstream in bokeh
# rbg = RadioButtonGroup(labels=list(md.columns),
# callback=callback)
toggle_buttons = [Toggle(label=x, callback=callback) for x in list(md.columns) + ["TPM", "FPKM"]]
else:
toggle_buttons = []
pca_components = sorted([int(x) + 1 for x in pca_source.column_names if re.match("\d+", x)])
menulist = ["{} ({:.2f}%)".format(x, 100.0 * p) for x, p in zip(pca_components, pcaobj.explained_variance_ratio_)]
component_x = Select(title = "PCA component x", options = menulist, value=menulist[0],
callback=xcallback)
component_y = Select(title = "PCA component y", options = menulist, value=menulist[1],
callback=ycallback)
# Make the pca plot
kwargs = {'plot_width': 400, 'plot_height': 400,
'title_text_font_size': "12pt",
'title': "Principal component analysis",
'tools': TOOLS,
'x_axis_label_text_font_size': '10pt',
'x_major_label_orientation': np.pi/3,
'y_axis_label_text_font_size': '10pt',
'y_major_label_orientation': np.pi/3,
}
fig = points('x', 'y', df=pca_source, color='color', size='size', alpha=0.7, **kwargs)
tooltiplist = [("sample", "@SM")] if "SM" in pca_source.column_names else []
if not md is None:
tooltiplist = tooltiplist + [(str(x), "@{}".format(x)) for x
in md.columns] + \
[("Detected genes (TPM)", "@TPM"), ("Detected genes (FPKM)", "@FPKM")]
tooltips(fig, HoverTool, tooltiplist)
# Loadings
loadings = pd.DataFrame(pcaobj.components_).T
loadings.columns = [str(x) for x in loadings.columns]
loadings['x'] = loadings['0']
loadings['y'] = loadings['1']
try:
loadings["gene_id"] = pcaobj.features
except:
smllogger.warn("failed to set gene_id")
raise
try:
loadings["gene_name"] = [pcaobj.labels[x] for x in loadings["gene_id"]]
except:
smllogger.warn("failed to set gene_name")
raise
loadings_source = ColumnDataSource(loadings)
kwargs.update({'title': "Loadings"})
loadings_fig = points(x='x', y='y', df=loadings_source,
**kwargs)
tooltips(loadings_fig, HoverTool, [('gene_id', '@gene_id'), ('gene_name', '@gene_name')])
x_loadings_callback = CustomJS(args=dict(source=loadings_source),
code="""pca_loadings(source, cb_obj, "x");""")
y_loadings_callback = CustomJS(args=dict(source=loadings_source),
code="""pca_loadings(source, cb_obj, "y");""")
menulist = ["{} ({:.2f}%)".format(x, 100.0 * p) for x, p in zip(pca_components, pcaobj.explained_variance_ratio_)]
loadings_component_x = Select(title = "PCA loading x", options = menulist, value=menulist[0],
callback=x_loadings_callback)
loadings_component_y = Select(title = "PCA loading y", options = menulist, value=menulist[1],
callback=y_loadings_callback)
# Add taptool url if present
if taptool_url:
loadings_fig.add_tools(TapTool(callback=OpenURL(url=taptool_url)))
# Detected genes, FPKM and TPM
kwargs.update({'title': 'Number of detected genes',
'xlabel': "Sample",
'ylabel': "Detected genes",
'x_range': list(pca_source.data["SM"])})
n_genes_fig = dotplot(df=pca_source, x="SM", y="TPM", **kwargs)
tooltips(n_genes_fig, HoverTool, [('sample', '@SM'),
('# genes (TPM)', '@TPM'),
('# genes (FPKM)', '@FPKM')])
buttons = toggle_buttons + [component_x, component_y] + [loadings_component_x, loadings_component_y]
return {'pca' : vform(*(buttons + [gridplot([[fig, loadings_fig, n_genes_fig]])]))}