In [1]:
import numpy as np
import scanpy as sc
import matplotlib.pyplot as plt
Load in the actual data and the generated data¶
We will use the loom files that were generated by seurat after processing the generated data.
In [2]:
# Validation or Test Data only
test_data = sc.read_loom('/Users/aliheydari/Box/SCIV-Material/IntegratedSplits/20K Brain Small/REAL.loom');
In [3]:
activa = sc.read_loom('/Users/aliheydari/Box/SCIV-Material/IntegratedSplits/20K Brain Small/ACTIVA.loom');
Now we want to get the UMAP and TSNE of each dataset (which was done in seurat. To see how it was done, please look at the R notebook in the same directory)
In [5]:
real_umap = test_data.obsm['umap_cell_embeddings'];
ACTIVA_umap = activa.obsm['umap_cell_embeddings'];
In [6]:
real_tsne = test_data.obsm['tsne_cell_embeddings'];
ACTIVA_tsne = activa.obsm['tsne_cell_embeddings'];
Now we want to get the UMAP and TSNE of each dataset (which was done in seurat. To see how it was done, please look at the R notebook in the same directory)
In [7]:
import time
start = time.time()
# Fixing random state for reproducibility
np.random.seed(19680801)
x_synth = ACTIVA_umap[:, 0]
y_synth = ACTIVA_umap[:, 1]
x_real= real_umap[:, 0]
y_real= real_umap[:, 1]
# definitions for the axes
left, width = 0.1, 0.65
bottom, height = 0.1, 0.65
spacing = 0.005
rect_scatter = [left, bottom, width, height]
rect_histx = [left, bottom + height + spacing, width, 0.2]
rect_histy = [left + width + spacing, bottom, 0.2, height]
# start with a rectangular Figure
plt.figure(figsize=(8, 8))
ax_scatter = plt.axes(rect_scatter)
ax_scatter.tick_params(direction='in', top=True, right=True)
ax_histx = plt.axes(rect_histx)
ax_histx.tick_params(direction='in', labelbottom=False)
ax_histy = plt.axes(rect_histy)
ax_histy.tick_params(direction='in', labelleft=False)
# the scatter plot:
synth_plt = ax_scatter.scatter(x_synth, y_synth, label="ACTIVA ", c="tomato")
real_plt = ax_scatter.scatter(x_real, y_real, label="Brain Small (Test Set)", c="deepskyblue")
# ax_scatter.legend()
synth_plt.set_alpha(1)
real_plt.set_alpha(0.4)
# now determine nice limits by hand:
binwidth = 0.25
lim = np.ceil(np.abs([x_real, y_real]).max() / binwidth) * binwidth
ax_scatter.set_xlim((-lim, lim + 2))
ax_scatter.set_ylim((-lim, lim))
bins = np.arange(-lim, lim + binwidth, binwidth)
synth_histx = ax_histx.hist(x_synth, bins=bins, alpha = 0.7, color="tomato")
synth_histy = ax_histy.hist(y_synth, bins=bins, orientation='horizontal',alpha = 0.7, color="tomato")
real_histx = ax_histx.hist(x_real, bins=bins, alpha = 0.5, color="deepskyblue")
real_histy = ax_histy.hist(y_real, bins=bins, alpha =0.5, orientation='horizontal', color="deepskyblue")
ax_histx.set_xlim(ax_scatter.get_xlim())
ax_histy.set_ylim(ax_scatter.get_ylim())
fig1 = plt.gcf()
plt.show()
print('Duration: {} seconds'.format(time.time() - start))
Duration: 0.6924591064453125 seconds
Now TSNE:
In [8]:
import time
start = time.time()
# Fixing random state for reproducibility
np.random.seed(19680801)
x_synth = ACTIVA_tsne[:, 0]
y_synth = ACTIVA_tsne[:, 1]
x_real= real_tsne[:, 0]
y_real= real_tsne[:, 1]
# definitions for the axes
left, width = 0.1, 0.65
bottom, height = 0.1, 0.65
spacing = 0.005
rect_scatter = [left, bottom, width, height]
rect_histx = [left, bottom + height + spacing, width, 0.2]
rect_histy = [left + width + spacing, bottom, 0.2, height]
# start with a rectangular Figure
plt.figure(figsize=(8, 8))
ax_scatter = plt.axes(rect_scatter)
ax_scatter.tick_params(direction='in', top=True, right=True)
ax_histx = plt.axes(rect_histx)
ax_histx.tick_params(direction='in', labelbottom=False)
ax_histy = plt.axes(rect_histy)
ax_histy.tick_params(direction='in', labelleft=False)
# the scatter plot:
synth_plt = ax_scatter.scatter(x_synth, y_synth, label="ACTIVA ", c="tomato")
real_plt = ax_scatter.scatter(x_real, y_real, label="Brain Small (Test Set)", c="deepskyblue")
# ax_scatter.legend()
synth_plt.set_alpha(1)
real_plt.set_alpha(0.4)
# now determine nice limits by hand:
binwidth = 0.25
lim = np.ceil(np.abs([x_real, y_real]).max() / binwidth) * binwidth
ax_scatter.set_xlim((-lim - 10, lim + 10))
ax_scatter.set_ylim((-lim - 10, lim + 10))
bins = np.arange(-lim, lim + binwidth, binwidth)
synth_histx = ax_histx.hist(x_synth, bins=bins, alpha = 0.7, color="tomato")
synth_histy = ax_histy.hist(y_synth, bins=bins, orientation='horizontal',alpha = 0.7, color="tomato")
real_histx = ax_histx.hist(x_real, bins=bins, alpha = 0.5, color="deepskyblue")
real_histy = ax_histy.hist(y_real, bins=bins, alpha =0.5, orientation='horizontal', color="deepskyblue")
ax_histx.set_xlim(ax_scatter.get_xlim())
ax_histy.set_ylim(ax_scatter.get_ylim())
fig1 = plt.gcf()
plt.show()
print('Duration: {} seconds'.format(time.time() - start))
Duration: 1.578930139541626 seconds
In [ ]: