def write_gene_pairs_names_in_each_cluster(run_name, gene_pair_indexs,
gene_pair_names,
cluster_labels_of_gene_pairs):
output_dir = os.path.join(FIGURE_DIR, run_name)
mkdirs([output_dir])
cluster_labels_of_gene_pairs = np.array(cluster_labels_of_gene_pairs)
gene_pairs = np.array(gene_pair_names)
# cluster_mapping = {cid: cluster for cid, cluster in enumerate(np.unique(cluster_labels_of_gene_pairs))}
Ncluster = len(np.unique(cluster_labels_of_gene_pairs))
for cluster_id in range(Ncluster):
gene_pairs_in_this_cluster = gene_pairs[cluster_labels_of_gene_pairs ==
cluster_id]
print(gene_pairs_in_this_cluster)
gene_pairs_indexs_in_this_cluster = gene_pair_indexs[
cluster_labels_of_gene_pairs == cluster_id]
print(gene_pairs_indexs_in_this_cluster)
out_arr = np.array([[
gene_pairs_indexs_in_this_cluster[idx],
gene_pair_name.split("_")[0],
gene_pair_name.split("_")[1]
] for idx, gene_pair_name in enumerate(gene_pairs_in_this_cluster)])
outdir = os.path.join(output_dir, "cluster_%d" % (cluster_id + 1))
mkdirs([outdir])
output_fp = os.path.join(
outdir, "gene_pair_names_of_cluster_%d.csv" % (cluster_id + 1))
np.savetxt(output_fp, out_arr[:, :], delimiter='\n', fmt="%s,%s,%s")
def get_unique_gene_pairs_names_in_each_cluster(run_name, gene_pairs,
cluster_labels_of_gene_pairs):
output_dir = os.path.join(GENE_PAIR_NAME_DATA, run_name)
#output_dir = os.path.join(FIGURE_DIR, run_name)
mkdirs([output_dir])
unique_gene_names = []
cluster_labels_of_gene_pairs = np.array(cluster_labels_of_gene_pairs)
gene_pairs = np.array(gene_pairs)
n_cluster = len(np.unique(cluster_labels_of_gene_pairs))
for cluster_id in range(n_cluster):
gene_pairs_in_this_cluster = gene_pairs[cluster_labels_of_gene_pairs ==
cluster_id]
all_gene_names_in_this_cluster = []
for gene_pair_name in gene_pairs_in_this_cluster:
for gene_name in gene_pair_name.split("_"):
all_gene_names_in_this_cluster.append(gene_name)
unique_gene_names_in_this_cluster = np.unique(
all_gene_names_in_this_cluster)
unique_gene_names.append(unique_gene_names_in_this_cluster)
output_fp = os.path.join(output_dir, "%d.csv" % (cluster_id + 1))
np.savetxt(output_fp,
np.array(unique_gene_names_in_this_cluster)[:],
delimiter='\n',
fmt="%s")
def fig8_plot_selected_cluster_trajectories(selected_clusters,
cluster_labels,
traj_lst,
run_name,
N_SAMPLE=4,
fig_format='png',
cmap="RdBu",
color_palette=None):
selected_clusters = [item - 1 for item in selected_clusters]
N_CLUSTER = len(selected_clusters)
plt.rc('axes', linewidth=2., labelsize=10.,
labelpad=6.) # length for x tick and padding
plt.rc('xtick.major', size=6, pad=3) # length for x tick and padding
plt.rc('ytick.major', size=6, pad=3) # length for y tick and padding
plt.rc('lines', mew=5, lw=4) # line 'marker edge width' and thickness
plt.rc('ytick', labelsize=8) # ytick label size
plt.rc('xtick', labelsize=8) # xtick label
plt.rc('figure', dpi=300) # Sets rendering resolution to 300
plt.rc('lines',
markersize=3.1) # Sets marker size for scatter to reasonable size
c_arr = np.array([(time_point + 1.) / 10. for time_point in range(10)])
for cid, cluster in enumerate(selected_clusters):
fig = plt.figure(1000 + cluster)
trajectories = np.array(traj_lst)[np.array(cluster_labels) == cluster]
# n_sample = N_SAMPLEx
# sampled_indexs = np.random.choice(np.arange(trajectories.shape[0]), min(trajectories.shape[0], n_sample),
# replace=False)
# sampled_trajectories = trajectories[sampled_indexs]
for traj in trajectories:
colorline(plt.gca(), traj[:, 0], traj[:, 1], c_arr, cmap=cmap)
plt.xlim(-17.14, 65)
plt.ylim(-22.23, 40)
plt.xlabel("PCA Component 1")
plt.ylabel("PCA Component 2")
# plt.title(titles_of_each_cluster[cid])
fig_dir = os.path.join(FIGURE_DIR, run_name)
mkdirs([fig_dir])
fig_fp = os.path.join(
fig_dir, "selected_cluster_%d.%s" % ((cluster + 1), fig_format))
plt.savefig(fig_fp,
transparent=True,
bbox_inches='tight',
pad_inches=0.1,
format='png')
def plot_hierarchical_cluster(df,
linkage,
color_palette,
distance_threshold,
label_arr,
link_cols,
run_name,
fig_format="png",
label_position=-400):
n_cluster = len(np.unique(label_arr))
row_colors = df.cluster_label.map(color_palette)
fig_dir = os.path.join(FIGURE_DIR, run_name)
mkdirs([fig_dir])
# cm = sns.clustermap(df, method="ward", col_cluster=True, col_colors=row_colors, yticklabels=True, figsize=(35, 35))
# fig_fp1 = os.path.join(fig_dir, "hierarchical_cluster_%d.%s" % (n_cluster, fig_format))
# cm.savefig(fig_fp1, dpi=200)
fig2 = plt.figure(2, figsize=(30, 30))
R = dendrogram(linkage,
no_labels=True,
leaf_rotation=90,
orientation="top",
leaf_font_size=8,
distance_sort='ascending',
color_threshold=distance_threshold,
above_threshold_color="black",
link_color_func=lambda x: link_cols[x])
prev_sum = 0
for lbl_id in range(n_cluster):
label_tmp = label_arr[label_arr == lbl_id].shape[0] * 10
coord = prev_sum + label_tmp * 0.3
prev_sum += label_tmp
plt.text(coord,
label_position,
"cluster %d" % (lbl_id + 1),
rotation=45,
backgroundcolor=color_palette[lbl_id])
plt.ylim([label_position - 20, linkage[-1, -2]])
fig_fp2 = os.path.join(
fig_dir,
"cluster_dendrogram_%d_with_leafs.%s" % (n_cluster, fig_format))
fig2.savefig(fig_fp2, dpi=200)
def plot_heatmap_serie_of_each_cluster(data_dict,
N_CLUSTER,
cluster_lst,
passed_pair_indexs,
run_name,
fig_format="png",
TARGET_CLUSTER_IDs=None):
'''
Plots given trajectories with a color that is specific for every trajectory's own cluster index.
Outlier trajectories which are specified with -1 in `cluster_lst` are plotted dashed with black color
'''
print(N_CLUSTER)
data_dict = data_dict[0]
N_COL = 10
Max_NROW = 5
TICKS = range(0, 21, 5)
N_SUBFIG_PER_FIG = Max_NROW * N_COL
cluster_lst = np.array(cluster_lst)
fig_dir = os.path.join(FIGURE_DIR, run_name)
mkdirs([fig_dir])
if TARGET_CLUSTER_IDs:
cluster_ids = TARGET_CLUSTER_IDs
else:
cluster_ids = range(N_CLUSTER)
for cluster in cluster_ids:
gene_pair_indexs = passed_pair_indexs[cluster_lst == cluster]
n_gene_pairs_in_cluster = len(gene_pair_indexs)
NFIG = int(math.ceil(float(n_gene_pairs_in_cluster) / Max_NROW))
sub_fig_dir = os.path.join(fig_dir, "cluster_%d" % (cluster + 1))
mkdirs([sub_fig_dir])
for i in range(NFIG):
if NFIG > 1:
fig_fp = os.path.join(
sub_fig_dir,
"cluster_%d_%d.%s" % (cluster + 1, i, fig_format))
else:
fig_fp = os.path.join(
sub_fig_dir, "cluster_%d.%s" % (cluster + 1, fig_format))
base_index = i * N_SUBFIG_PER_FIG
N_remaining_files = n_gene_pairs_in_cluster * N_COL - base_index
N_ROW = int(math.ceil(
float(N_remaining_files) /
N_COL)) if N_remaining_files <= N_SUBFIG_PER_FIG else Max_NROW
fig, axs = plt.subplots(N_ROW,
N_COL,
figsize=(N_COL * EACH_SUB_FIG_SIZE,
N_ROW * EACH_SUB_FIG_SIZE))
SUB_FIG_RANGE = N_SUBFIG_PER_FIG if N_remaining_files > N_SUBFIG_PER_FIG else N_remaining_files
plt.set_cmap('viridis_r')
for j in range(SUB_FIG_RANGE):
row = j // N_COL
col = j % N_COL
if N_ROW == 1:
ax = axs[col]
else:
ax = axs[row][col]
gene_pair_id = gene_pair_indexs[i * Max_NROW + row] + 1
prob2d_array = data_dict[gene_pair_id][
'prob2d'] # shape 21* 21
prob2d = prob2d_array[col]
q_potential = -np.log(np.abs(prob2d))
cax = ax.pcolormesh(q_potential, vmin=3, vmax=14)
ax.set_yticks(TICKS)
ax.set_xticks(TICKS)
if row == 0:
ax.set_title("Stage %d" % Stages[col])
if col == 0:
ax.set_ylabel(data_dict[gene_pair_id]['pair_name'])
if j == 0:
fig.colorbar(cax, ax=ax)
plt.savefig(fig_fp, dpi=200)
print("cluster %d" % (cluster + 1))
def plot_cluster(traj_lst,
cluster_lst,
run_name,
fig_format="png",
color_palette=None,
log_transformed=True,
cmap="gist_rainbow"):
'''
Plots given trajectories with a color that is specific for every trajectory's own cluster index.
Outlier trajectories which are specified with -1 in `cluster_lst` are plotted dashed with black color
'''
if log_transformed:
X_MAX_LIM = 65
Y_MAX_LIM = 40
X_MIN_LIM = -17.14
Y_MIN_LIM = -22.23
else:
X_MAX_LIM = 0.6
Y_MAX_LIM = 0.15
X_MIN_LIM = -0.2
Y_MIN_LIM = -0.1
N_CLUSTER = len(np.unique(cluster_lst))
cluster_mapping = {
cluster: cid
for cid, cluster in enumerate(np.unique(cluster_lst))
}
N_COL = 4
N_ROW = int(math.ceil(float(N_CLUSTER) / N_COL))
c_arr = np.array([(time_point + 1.) / 10. for time_point in range(10)])
traj_lst = np.array(traj_lst)
cluster_lst = np.array(cluster_lst)
fig, axs = plt.subplots(N_ROW,
N_COL,
figsize=(N_COL * EACH_SUB_FIG_SIZE,
N_ROW * EACH_SUB_FIG_SIZE))
for index, (traj, cluster) in enumerate(zip(traj_lst, cluster_lst)):
row = cluster_mapping[cluster] // N_COL
col = cluster_mapping[cluster] % N_COL
if N_ROW > 1:
ax = axs[row][col]
else:
ax = axs[col]
colorline(ax, traj[:, 0], traj[:, 1], c_arr, cmap=cmap)
ax.set_xlim(X_MIN_LIM, X_MAX_LIM)
ax.set_ylim(Y_MIN_LIM, Y_MAX_LIM)
if row == N_ROW - 1:
ax.set_xlabel("PCA Component 1")
if col == 0:
ax.set_ylabel("PCA Component 2")
if color_palette:
ax.set_title(
"cluster %d" % (cluster_mapping[cluster] + 1),
backgroundcolor=color_palette[cluster_mapping[cluster]])
else:
ax.set_title("cluster %d" % (cluster_mapping[cluster] + 1))
fig_dir = os.path.join(FIGURE_DIR, run_name)
mkdirs([fig_dir])
fig_fp = os.path.join(
fig_dir, "trajactory_clusters_%d.%s" % (N_CLUSTER, fig_format))
plt.savefig(fig_fp,
transparent=True,
bbox_inches='tight',
pad_inches=0.1,
format='png')
#plt.show()
return [cluster_mapping[cluster] for cluster in cluster_lst]
from matplotlib import collections as mcoll
from Util import mkdirs
#%matplotlib inline
# Utility Functions
color_lst = plt.rcParams['axes.prop_cycle'].by_key()['color']
color_lst.extend([
'firebrick', 'olive', 'indigo', 'khaki', 'teal', 'saddlebrown', 'skyblue',
'coral', 'darkorange', 'lime', 'darkorchid', 'dimgray'
])
DATA_DIR = os.path.join(os.path.abspath(os.curdir), 'DATA')
PICKLE_DATA = os.path.join(DATA_DIR, "pickle_data")
NPY_DATA = os.path.join(DATA_DIR, "npy_data")
GENE_PAIR_NAME_DATA = os.path.join(DATA_DIR, "gene_pair_names")
FIGURE_DIR = os.path.join(os.path.abspath(os.curdir), 'Figures')
mkdirs([PICKLE_DATA, NPY_DATA, FIGURE_DIR, GENE_PAIR_NAME_DATA])
Stages = [8, 10, 11, 12, 13, 14, 16, 18, 20, 22]
EACH_SUB_FIG_SIZE = 5
FIGURE_FORMAT = "png"
def make_segments(x, y):
"""
Create list of line segments from x and y coordinates, in the correct format
for LineCollection: an array of the form numlines x (points per line) x 2 (x
and y) array
"""
points = np.array([x, y]).T.reshape(-1, 1, 2)
segments = np.concatenate([points[:-1], points[1:]], axis=1)
return segments