Skip to content

Commit

Permalink
mod figure 1
Browse files Browse the repository at this point in the history
  • Loading branch information
@MohamedOmar2020
MohamedOmar2020 committed Aug 16, 2023
1 parent d5aea23 commit f918f67
Show file tree
Hide file tree
Showing 12 changed files with 1,087 additions and 526 deletions.
110 changes: 5 additions & 105 deletions code/Figure1_S1.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -89,109 +89,6 @@
###########################
# 1B
###########################
# umap by n_genes
sc.pl.umap(
adata_mouse_mesenchyme,
color="n_genes", size = 10,
title = '', cmap = 'viridis',
save = '_mouse_genes'
)

# umap by n_counts
sc.pl.umap(
adata_mouse_mesenchyme,
color="n_counts", size = 10,
title = '', cmap = 'viridis',
save = '_mouse_counts'
)

###########################
# 1C
###########################
## heatmap of fraction of cells of each stromal cluster by mouse models
def distribution(
adata,
partition="cluster",
labels="key_new",
modelorder=['B6', 'B6.129', 'FVBN', 'WT for PN', 'WT for PRN', 'T-ERG', 'NP', 'Hi-MYC', 'PRN'],
partitionorder=None,
figsize=(10, 7),
):
"""
plots for distribution of cells by key labels
"""

models = []
for name in adata.obs[labels].cat.categories.tolist():
model = adata[adata.obs[labels] == name]
models.append((name, model))

# generate percentage dict by model
bymodeldict = {}
for name, model in models:
total = len(model.obs)
modeldict = {}
for j in adata.obs[partition].values.categories:
modeldict[j] = np.sum(model.obs[partition] == j) / total
bymodeldict[name] = modeldict

# subset wt and mut
mutant = adata[adata.obs["condition"] == "mutant"]
wildtype = adata[adata.obs["condition"] == "wildtype"]
mutwt = [("mutant", mutant), ("wildtype", wildtype)]

# compute composition of partition by keys
byclusterdict = {}
for cluster in adata.obs[partition].cat.categories:
clusterdata = adata[adata.obs[partition] == cluster]
total = np.sum(adata.obs[partition] == cluster)
clusterdict = {}
for key in adata.obs[labels].cat.categories.tolist():
clusterdict[key] = np.sum(clusterdata.obs[labels] == key) / total
byclusterdict[cluster] = clusterdict
byclusterdf = pd.DataFrame.from_dict(byclusterdict, orient="index")
byclusterdf.to_csv("modelcellcomposition2.csv")

# generate percentage dict by mutwt
bymutwtdict = {}
for name, model in mutwt:
total = len(model.obs)
modeldict = {}
for j in adata.obs[partition].values.categories:
modeldict[j] = np.sum(model.obs[partition] == j) / total
bymutwtdict[name] = modeldict

# generate df wildtype/mutant
lollipopdict = {}
for i in adata.obs[partition].values.categories:
lollipopdict[i] = bymutwtdict["wildtype"][i] / bymutwtdict["mutant"][i]
lollipopdf = pd.DataFrame.from_dict(lollipopdict, orient="index")
lollipopdf = lollipopdf.rename(columns={0: "ratio of percentage (wt/mut)"})
if partitionorder:
lollipopdf = lollipopdf.reindex(partitionorder)

# plot heatmapall
import seaborn as sns

sns.set(rc={"figure.figsize": figsize})
bymodeldf = pd.DataFrame.from_dict(bymodeldict, orient="index")
if modelorder:
bymodeldf = bymodeldf.reindex(modelorder)
if partitionorder:
bymodeldf = bymodeldf[partitionorder]
bymodeldf.to_csv("modelcellcomposition.csv")
ax = sns.heatmap(
bymodeldf, cmap="RdBu_r", center=0.05, annot=True, fmt=".1%", cbar=False
)
ax.set_yticklabels(ax.get_yticklabels(), rotation = 0)
ax.get_figure().savefig(f"percentages.png", dpi=500, bbox_inches="tight")


distribution(adata_mouse_mesenchyme)

###########################
# 1D
###########################
# parallel categories mouse models

# Arrange the cell annotation dataframe by the clusters
Expand All @@ -213,6 +110,9 @@ def distribution(
fig.write_image("figures/figures_cell/parallel_categories_cluster.png", scale = 2)

###############################
# 1E
# 1C-D
###############################
# refer to the R script
# refer to the R script


#############################################
108 changes: 107 additions & 1 deletion code/Figure2_S2.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -155,7 +155,113 @@
# Figure S2
###############################

# scenic heatmap
###########################
# S2A
###########################
# umap by n_genes
sc.pl.umap(
adata_mouse_mesenchyme,
color="n_genes", size = 10,
title = '', cmap = 'viridis',
save = '_mouse_genes'
)

# umap by n_counts
sc.pl.umap(
adata_mouse_mesenchyme,
color="n_counts", size = 10,
title = '', cmap = 'viridis',
save = '_mouse_counts'
)


###########################
# S2B
###########################
## heatmap of fraction of cells of each stromal cluster by mouse models
def distribution(
adata,
partition="cluster",
labels="key_new",
modelorder=['B6', 'B6.129', 'FVBN', 'WT for NP', 'WT for PRN', 'T-ERG', 'NP', 'Hi-MYC', 'PRN'],
partitionorder=None,
figsize=(10, 7),
):
"""
plots for distribution of cells by key labels
"""

models = []
for name in adata.obs[labels].cat.categories.tolist():
model = adata[adata.obs[labels] == name]
models.append((name, model))

# generate percentage dict by model
bymodeldict = {}
for name, model in models:
total = len(model.obs)
modeldict = {}
for j in adata.obs[partition].values.categories:
modeldict[j] = np.sum(model.obs[partition] == j) / total
bymodeldict[name] = modeldict

# subset wt and mut
mutant = adata[adata.obs["condition"] == "mutant"]
wildtype = adata[adata.obs["condition"] == "wildtype"]
mutwt = [("mutant", mutant), ("wildtype", wildtype)]

# compute composition of partition by keys
byclusterdict = {}
for cluster in adata.obs[partition].cat.categories:
clusterdata = adata[adata.obs[partition] == cluster]
total = np.sum(adata.obs[partition] == cluster)
clusterdict = {}
for key in adata.obs[labels].cat.categories.tolist():
clusterdict[key] = np.sum(clusterdata.obs[labels] == key) / total
byclusterdict[cluster] = clusterdict
byclusterdf = pd.DataFrame.from_dict(byclusterdict, orient="index")
byclusterdf.to_csv("modelcellcomposition2.csv")

# generate percentage dict by mutwt
bymutwtdict = {}
for name, model in mutwt:
total = len(model.obs)
modeldict = {}
for j in adata.obs[partition].values.categories:
modeldict[j] = np.sum(model.obs[partition] == j) / total
bymutwtdict[name] = modeldict

# generate df wildtype/mutant
lollipopdict = {}
for i in adata.obs[partition].values.categories:
lollipopdict[i] = bymutwtdict["wildtype"][i] / bymutwtdict["mutant"][i]
lollipopdf = pd.DataFrame.from_dict(lollipopdict, orient="index")
lollipopdf = lollipopdf.rename(columns={0: "ratio of percentage (wt/mut)"})
if partitionorder:
lollipopdf = lollipopdf.reindex(partitionorder)

# plot heatmapall
import seaborn as sns

sns.set(rc={"figure.figsize": figsize})
bymodeldf = pd.DataFrame.from_dict(bymodeldict, orient="index")
if modelorder:
bymodeldf = bymodeldf.reindex(modelorder)
if partitionorder:
bymodeldf = bymodeldf[partitionorder]
bymodeldf.to_csv("modelcellcomposition.csv")
ax = sns.heatmap(
bymodeldf, cmap="RdBu_r", center=0.05, annot=True, fmt=".1%", cbar=False
)
ax.set_yticklabels(ax.get_yticklabels(), rotation = 0)
ax.get_figure().savefig(f"percentages.png", dpi=500, bbox_inches="tight")


distribution(adata_mouse_mesenchyme)

###########################
# S2C: scenic heatmap
###########################
# load the scenic output
auc_mtx = pd.read_csv('data/for_mouse/scenic_mesenchyme_auc.csv', index_col=0)
bin_mtx = pd.read_csv('data/for_mouse/scenic_mesenchyme_binary.csv', index_col=0)
Expand Down
Loading

0 comments on commit f918f67

Please sign in to comment.