1、cellphoneBD Rstudio 前期准备工作

建立输入文件
setwd("D:/D/metropolypus/FastMNN/cellphoneDB")
###根据colname名字区分BD单细胞和10X单细胞测序，提取10X单细胞测序数据
colnames(sce)
sce$tech <- sapply(colnames(x = sce), function(x) unlist(strsplit(x, "_"))[1]) 
sce$IF=sapply(sce$tech, function(x) unlist(nchar(x)>10))
table(sce$IF)

Idents(sce)<-sce$IF
ten=subset(sce,idents = c("TRUE"))
table(ten$sample)

subcell <- sample(colnames(ten),10000)
sce <- ten[,subcell]

library(SeuratDisk)
# 将Seurat对象保存为h5Seurat文件
SaveH5Seurat(sce, filename = "seurat_obj.h5Seurat")
# 转换为h5ad格式
Convert("seurat_obj.h5Seurat", dest = "h5ad")


# 安装pandas和anndata
# pip install pandas anndata
import pandas as pd
import anndata
# 读取CSV文件
expression_matrix = pd.read_csv("/mnt/d/D/temp/cellphoneDB/expression_matrix.csv", index_col=0)
# 创建AnnData对象
adata = anndata.AnnData(X=expression_matrix)
# 保存为.h5ad文件
adata.write("/mnt/d/D/temp/cellphoneDB/sce.h5ad")
print(count_cells[:10])

sce$Cell = rownames(sce@meta.data)
df = sce@meta.data[,c("Cell","celltype")]
colnames(df) = c("Cell","cell_type")
df$Cell <- gsub("-", ".", df$Cell)
write.table(df, file = "meta1.txt",sep = "\t",quote = F,row.names = F)

sce$Cell = rownames(sce@meta.data)
df = sce@meta.data[,c("celltype","group")]
colnames(df) = c("cell_type","microenvironment")
write.table(df, file = "microenvironment.txt",sep = "\t",quote = F,row.names = F)

DEG.Au <-dplyr::filter(DEG.A,DEG.A$avg_log2F >0.25)
write.table(DEG.Au, file = "degs_file.txt",sep = "\t",quote = F,row.names = F)

2、CellPhoneDB python中运行

#https://github.com/ventolab/CellphoneDB
#https://github.com/ventolab/CellphoneDB/blob/master/NatureProtocols2024_case_studies/CaseExample1_differentiation/analysis_method3_CellSign_microenvironments.ipynb
conda create -n cpdb python=3.8
source activate
conda activate cpdb
source activate cpdb
cd /mnt/d/D/temp/cellphoneDB

import pandas as pd
# 加载 meta 文件
meta_df = pd.read_csv("/mnt/d/D/temp/cellphoneDB/meta1.txt", sep="\t")
meta_clusters = set(meta_df['cell_type'])  # 使用meta文件中实际使用的列名

# 加载 DEGs 文件（需要早excel表格中手动调节DEGs每列的顺序需要和example_data中DEGs的顺序保持一致）
degs_df = pd.read_csv("/mnt/d/D/temp/cellphoneDB/degs_file.txt", sep="\t") 
degs_clusters = set(degs_df.iloc[:, 5])  # 取DEGs文件的第一列

# 检查 DEGs 中不存在于 meta 中的集群/细胞类型
missing_clusters = degs_clusters - meta_clusters
if missing_clusters:
    print("Missing clusters in meta:", missing_clusters)
else:
    print("All clusters match.")


from cellphonedb.src.core.methods import cpdb_degs_analysis_method
cpdb_results = cpdb_degs_analysis_method.call(
         cpdb_file_path = "/mnt/d/D/temp/cellphoneDB/cellphonedb.zip",
         meta_file_path = "/mnt/d/D/temp/cellphoneDB/meta1.txt",
         counts_file_path = "/mnt/d/D/temp/cellphoneDB/sce.h5ad",
         degs_file_path = "/mnt/d/D/temp/cellphoneDB/degs_file.txt",
         counts_data = 'hgnc_symbol',
         threshold = 0.1,
         output_path = "/mnt/d/D/temp/cellphoneDB/out/")

##代码运行时，所有表格列名的位置都要和D:\D\temp\cellphonedb\CellphoneDB-master\CellphoneDB-master\example_data中顺序一样

import anndata
adata = anndata.read_h5ad("/mnt/d/D/temp/cellphoneDB/seurat_obj.h5ad")
import matplotlib.pyplot as plt

# 调用绘制热图的函数
kpy.plot_cpdb_heatmap(
    pvals=cpdb_results['relevant_interactions'],
    degs_analysis=True,
    figsize=(5, 5),
    title="Sum of significant interactions"
)

# 保存图像到指定路径
output_path = "/mnt/d/D/temp/cellphoneDB/heatmap.png"
plt.savefig(output_path, format='png', dpi=300, bbox_inches='tight')

# 关闭当前图以释放内存
plt.close()


from plotnine import facet_wrap
# 创建图形对象
p = kpy.plot_cpdb(
    adata=adata,
    cell_type1=".",
    cell_type2="Mast_cells",
    means=cpdb_results['means'],
    pvals=cpdb_results['relevant_interactions'],
    celltype_key="celltype",
    figsize=(12, 8),
    title="Interactions with Mast cells",
    max_size=6,
    highlight_size=0.75,
    degs_analysis=True,
    standard_scale=False
)
# 添加facet_wrap
final_plot = p + facet_wrap("~ classification", ncol=2)
# 使用 plotnine 的 draw 方法，然后调用保存
output_path = "/mnt/d/D/temp/cellphoneDB/interaction_plot.mastcells.pdf"
from plotnine import ggplot
final_plot.save(filename=output_path, format='pdf', dpi=600, height=20, width=12, units='in')

p=kpy.plot_cpdb(
    adata = adata,
    cell_type1 = "Mast_cells",
    cell_type2 = ".", 
    means = cpdb_results['means'],
    pvals = cpdb_results['relevant_interactions'],
    celltype_key = "celltype",
#     genes = ["JAG1"],
    figsize = (10,15),
    title = "Interactions between oocytes and granulosa in the follicles",
    max_size = 5,
    highlight_size = 0.75,
    degs_analysis = True,
    standard_scale = False,
#     interaction_scores = cpdb_results['interaction_scores'],
    scale_alpha_by_interaction_scores=True,
)
output_path = "/mnt/d/D/temp/cellphoneDB/interaction_mastcell.allcells1.pdf"
from plotnine import ggplot
p.save(filename=output_path, format='pdf', dpi=600, height=8, width=8, units='in')
p.save(filename=output_path, format='pdf', dpi=300)
# 关闭当前图以释放内存
plt.close()

p = kpy.plot_cpdb(
    adata = adata,
    cell_type1 = ".",
    cell_type2 = "Mast_cells", 
    means = cpdb_results['means'],
    pvals = cpdb_results['relevant_interactions'],
    celltype_key = "celltype",
    genes = ["DHRS9"],
    figsize = (10,5),
    title = "Interactions involving DHRS9",
    max_size = 5,
    highlight_size = 0.75,
    degs_analysis = True,
    standard_scale = False
)
# 使用 plotnine 的 draw 方法，然后调用保存
output_path = "/mnt/d/D/temp/cellphoneDB/interaction_plotDHRS9-PGR.pdf"
final_plot.save(filename=output_path, format='pdf', dpi=300)

# 关闭当前图以释放内存
plt.close()
p = kpy.plot_cpdb(
    adata = adata,
    cell_type1 = "unciliated_EpC|ciliated_EpC",
    cell_type2 = "Mast_cells", 
    means = cpdb_results['means'],
    pvals = cpdb_results['relevant_interactions'],
    celltype_key = "celltype",
    figsize = (8, 5),
    title = "Interactions between oocytes and granulosa in the follicles \n supported by downstream TF activity",
    max_size = 6,
    highlight_size = 0.75,
    degs_analysis = True,
    standard_scale = False,
    cellsign = cpdb_results['CellSign_active_interactions'],
    filter_by_cellsign = True,
    scale_alpha_by_cellsign = False
)