【视频教程】Cellphonedb v5更新：python版Sc

之前我们演示cpdb分析的时候侧重的是R分析seurat对象准备input data，小伙伴提到“我单细胞一开始就是用scanpy分析的，怎么进行CPDB分析呢？”，其实之前的cellphonedb V5教程里面我们已经提到了。然而我们前面的帖子也分享了h5ad转seurat的方式（玩转单细胞(16)：Scanpy单细胞h5ad数据转化为Seurat对象），以及seurat转h5ad的多个方式，所以他们之间不存在隔阂了，搞不定互相转化就好了。
cpdb分析counts_file_path参数可以是表达矩阵，也可以是单细胞分析结果h5ad对象。但是矩阵是normalize后的，我们演示使用的数据还是scanpy官网教程中PBMC的数据，数据没有任何生物学意义，仅仅作为cpdb流程演示。

import pandas as pd
import scanpy as sc
import numpy as np

#load data
adata = sc.read_h5ad('./pbmc3k.h5ad')
adata = adata.raw.to_adata()
adata
adata_cpdb = adata[~adata.obs['leiden'].isin(['Megakaryocytes'])]
#normalize
counts = adata_cpdb.X.toarray()
adata_cpdb.X = counts
sc.pp.normalize_per_cell(adata_cpdb, counts_per_cell_after=1e4)
adata_cpdb.layers["norm"] = adata_cpdb.X
counts = pd.DataFrame(data=adata_cpdb.layers["norm"], index=adata_cpdb.obs.index.tolist(), columns=adata_cpdb.var.index.tolist()).T
counts.index.name = "Gene"
#save meta
meta = pd.DataFrame(data=adata_cpdb.obs['leiden'].tolist(), index=adata_cpdb.obs.index.tolist(), columns=["Cell type"])
meta.index.name = "Cell"
meta

CDPB analysis:

import numpy as np
import pandas as pd
import scanpy as sc
import anndata
import os
import sys
from scipy import sparse
#Download database from source
from IPython.display import HTML, display
from cellphonedb.utils import db_releases_utils
display(HTML(db_releases_utils.get_remote_database_versions_html()['db_releases_html_table']))
cpdb_version = 'v5.0.0'
cpdb_target_dir = os.path.join('./', cpdb_version)

from cellphonedb.utils import db_utils
db_utils.download_database(cpdb_target_dir, cpdb_version)
sc.settings.verbosity = 3  # verbosity: errors (0), warnings (1), info (2), hints (3)
sc.logging.print_versions()
sc.settings.set_figure_params(dpi=100)  # low dpi (dots per inch) yields small inline figures
cpdb_file_path = './v5.0.0/cellphonedb.zip'
meta_file_path = './meta_cpdb.txt'
counts_file_path = './count_cpdb.txt'
out_path = './test1'
from cellphonedb.src.core.methods import cpdb_statistical_analysis_method

cpdb_results1 = cpdb_statistical_analysis_method.call(
    cpdb_file_path = cpdb_file_path,               
    meta_file_path = meta_file_path,                
    counts_file_path = counts_file_path,           
    counts_data = 'hgnc_symbol',                    
    #active_tfs_file_path = active_tf_path,          
    #microenvs_file_path = microenvs_file_path,      
    score_interactions = True,                      
    iterations = 1000,                              
    threshold = 0.1,                                 
    threads = 5,                                    
    debug_seed = 42,                                
    result_precision = 3,                           
    pvalue = 0.05,                                   
    separator = '|',                                 
    debug = False,                                  
    output_path = out_path                        
    )

CPDB分析：input h5ad & meta！

cpdb_file_path = './v5.0.0/cellphonedb.zip'
meta_file_path = './meta_cpdb.txt'
counts_file_path = './adata_cpdb.h5ad'
out_path = './test2'
from cellphonedb.src.core.methods import cpdb_statistical_analysis_method

cpdb_results2 = cpdb_statistical_analysis_method.call(
    cpdb_file_path = cpdb_file_path,               
    meta_file_path = meta_file_path,                
    counts_file_path = counts_file_path,           
    counts_data = 'hgnc_symbol',                    
    #active_tfs_file_path = active_tf_path,          
    #microenvs_file_path = microenvs_file_path,      
    score_interactions = True,                      
    iterations = 1000,                              
    threshold = 0.1,                                 
    threads = 5,                                    
    debug_seed = 42,                                
    result_precision = 3,                           
    pvalue = 0.05,                                   
    separator = '|',                                 
    debug = False,                                  
    output_path = out_path                        
    )

整合结果：两种方式一样的结果！这种结果可以导出在R或者py中进行各种可视化！

cpdb_pbmc = summary_cpdb_results(cpdb_results1)
cpdb_pbmc.head()

利用cpdb里面的函数，提取结果进行一些可视化！

from cellphonedb.utils import search_utils

search_results = search_utils.search_analysis_results(
    query_cell_types_1 = ['CD4 T'],  # List of cells 1, will be paired to cells 2 (list or 'All').
    query_cell_types_2 = ['CD14 Monocytes', 'FCGR3A Monocytes'],     # List of cells 2, will be paired to cells 1 (list or 'All').
    # query_genes = ['TGFBR1'],                                       # filter interactions based on the genes participating (list).
    #query_interactions = ['CSF1_CSF1R'],                            # filter intereactions based on their name (list).
    significant_means = cpdb_results1['significant_means'],          # significant_means file generated by CellphoneDB.
    deconvoluted = cpdb_results1['deconvoluted'],                    # devonvoluted file generated by CellphoneDB.
    interaction_scores = cpdb_results1['interaction_scores'],        # interaction score generated by CellphoneDB.
    query_minimum_score = 50,                                       # minimum score that an interaction must have to be filtered.
    separator = '|',                                                # separator (default: |) employed to split cells (cellA|cellB).
    long_format = True                                            # converts the output into a wide table, removing non-significant interactions
)

search_results.head()