def plot_pca(dat,
colour_subgroups,
p=None,
components=(0, 1),
marker_subgroups=None,
ax=None,
colour_map=None,
marker_map=None,
**kwargs):
if p is None:
p = pca.PCA()
pca_data = p.fit_transform(dat.transpose())
else:
pca_data = p.transform(dat.transpose())
variance_explained = p.explained_variance_ratio_ * 100.
ax = scatter.scatter_with_colour_and_markers(
pca_data[:, components],
colour_subgroups=colour_subgroups,
colour_map=colour_map,
marker_subgroups=marker_subgroups,
marker_map=marker_map,
ax=ax,
**kwargs)
ax.set_xlabel("PCA component %s (%.1f%%)" %
(components[0] + 1, variance_explained[components[0]]))
ax.set_ylabel("PCA component %s (%.1f%%)" %
(components[1] + 1, variance_explained[components[1]]))
return p, ax
def plot_pca(dat,
colour_subgroups,
p=None,
components=(0, 1),
marker_subgroups=None,
ax=None,
colour_map=None,
marker_map=None,
**kwargs):
if p is None:
p = pca.PCA()
pca_data = p.fit_transform(dat.transpose())
else:
pca_data = p.transform(dat.transpose())
variance_explained = p.explained_variance_ratio_ * 100.
ax = scatter.scatter_with_colour_and_markers(
pca_data[:, components],
colour_subgroups=colour_subgroups,
colour_map=colour_map,
marker_subgroups=marker_subgroups,
marker_map=marker_map,
ax=ax,
**kwargs)
# Update: from Seaborn v 0.9.0, the legend appears with a frame by default, but this doesn't look good.
leg = ax.get_legend()
leg.set_frame_on(False)
ax.set_xlabel("PCA component %s (%.1f%%)" %
(components[0] + 1, variance_explained[components[0]]))
ax.set_ylabel("PCA component %s (%.1f%%)" %
(components[1] + 1, variance_explained[components[1]]))
return p, ax
def compare_two_gene_levels(
dat_two_cols,
meta,
legend_dict,
colour_map=scatter_colours,
marker_map=scatter_markers,
):
ax = scatter.scatter_with_colour_and_markers(
dat_two_cols,
colour_subgroups=meta.patient_id,
colour_map=colour_map,
marker_subgroups=meta.type,
marker_map=marker_map
)
common.add_custom_legend(ax, legend_dict, loc_outside=True)
ax.set_xlabel('%s (logTPM)' % dat_two_cols.columns[0])
ax.set_ylabel('%s (logTPM)' % dat_two_cols.columns[1])
fig = ax.figure
fig.tight_layout()
fig.subplots_adjust(right=0.8)
return fig, ax
cmap = collections.OrderedDict([('fl/fl', 'k'),
(r'$\Delta$/$\Delta$', 'w')])
marker_subgroups = colour_subgroups
mmap = collections.OrderedDict([('fl/fl', 'o'),
(r'$\Delta$/$\Delta$', 's')])
p = PCA()
pc_dat = p.fit_transform(log_dat.transpose())
variance_explained = p.explained_variance_ratio_ * 100.
ax = scatter.scatter_with_colour_and_markers(
pc_dat[:, [0, 1]],
colour_subgroups=None,
colour_map=None,
marker_subgroups=marker_subgroups,
marker_map=mmap,
default_colour='k',
ms=50)
ax.set_xlabel("PCA component %s (%.1f%%)" % (1, variance_explained[0]),
fontsize=fontsize)
ax.set_ylabel("PCA component %s (%.1f%%)" % (2, variance_explained[1]),
fontsize=fontsize)
leg = ax.legend()
fr = leg.get_frame()
fr.set_edgecolor('k')
plt.setp(leg.get_texts(), fontsize=fontsize)
plt.setp(ax.xaxis.get_ticklabels(), fontsize=fontsize)
plt.setp(ax.yaxis.get_ticklabels(), fontsize=fontsize)
"all_samples_pca_019_only_annotated.png"),
dpi=200)
# Again, again! But this time we compute the PCs with only 019 lines, then transform the remaining lines
# using the same PCs
p = PCA()
pca_data = p.fit_transform(mdat_019.transpose())
variance_explained = p.explained_variance_ratio_ * 100.
fig = plt.figure(figsize=(6.4, 4.8))
ax = fig.add_subplot(111)
ax = scatter.scatter_with_colour_and_markers(
pca_data[:, [0, 1]],
colour_subgroups=obj.meta.loc[mdat_019.columns, 'patient_id'],
colour_map=cmap,
marker_subgroups=obj.meta.loc[mdat_019.columns, 'descriptor'],
marker_map=mmap,
ax=ax,
)
ax.set_xlabel("PCA component %s (%.1f%%)" % (1, variance_explained[0]))
ax.set_ylabel("PCA component %s (%.1f%%)" % (2, variance_explained[1]))
leg = ax.get_legend()
leg.set_frame_on(False)
fig.subplots_adjust(left=0.12, right=0.75, top=0.98)
fig.savefig(os.path.join(outdir, "pca_019_only.png"), dpi=200)
for i, col in enumerate(mdat_019.columns):
this_coords = pca_data[i, :2]
ax.annotate(col, this_coords)
