{"id":60891,"date":"2024-09-12T16:02:30","date_gmt":"2024-09-12T08:02:30","guid":{"rendered":"http:\/\/www.biocloudservice.com\/wordpress\/?p=60891"},"modified":"2024-09-12T16:02:30","modified_gmt":"2024-09-12T08:02:30","slug":"%e7%bb%88%e4%ba%8e%e6%95%b4%e7%90%86%e5%a5%bd%e4%ba%86%ef%bc%81%e7%bb%93%e5%90%88%e7%89%b9%e5%be%81%e7%9a%84%e5%8d%95%e7%bb%86%e8%83%9e%e5%b5%8c%e5%85%a5r%e5%8c%85simba%e5%85%a8%e5%a5%97%e5%88%86","status":"publish","type":"post","link":"http:\/\/www.biocloudservice.com\/wordpress\/?p=60891","title":{"rendered":"\u7ec8\u4e8e\u6574\u7406\u597d\u4e86\uff01\u7ed3\u5408\u7279\u5f81\u7684\u5355\u7ec6\u80de\u5d4c\u5165R\u5305SIMBA\u5168\u5957\u5206\u6790\u6559\u7a0b\uff0b\u4ee3\u7801\u9001\u7ed9\u5927\u5bb6"},"content":{"rendered":"\n

\u524d\u8a00<\/strong><\/strong><\/p>\n\n\n\n

\u8c08\u5230\u5355\u7ec6\u80de\u5206\u6790\uff0c\u5c11\u4e0d\u4e86\u7684\u5c31\u662f\u805a\u7c7b\uff01\u805a\u7c7b\uff01\u5728\u805a\u7c7b\uff01\uff01\u4e0d\u5e78\u7684\u662f\uff0c\u8fd9\u4e9b\u201c\u4ee5\u805a\u7c7b\u4e2d\u5fc3\u201d\u5206\u6790\u65b9\u6cd5\u4f9d\u8d56\u4e8e\u51c6\u786e\u5b9a\u4e49\u7684\u805a\u7c7b\u89e3\u51b3\u65b9\u6848\u6765\u53d1\u73b0\u6709\u610f\u4e49\u7684\u6807\u8bb0\u7279\u5f81\u3002\u7531\u4e8e\u805a\u7c7b\u89e3\u51b3\u65b9\u6848\u5728\u7528\u6237\u5b9a\u4e49\u7684\u805a\u7c7b\u5206\u8fa8\u7387\uff08\u805a\u7c7b\u6570\uff09\u548c\u9009\u62e9\u7684\u805a\u7c7b\u7b97\u6cd5\u7a7a\u95f4\u5185\u53ef\u80fd\u6709\u5f88\u5927\u7684\u53d8\u5316\u8303\u56f4\uff0c\u8fd9\u53ef\u80fd\u5bfc\u81f4\u4e0d\u4e00\u81f4\u548c\u4e0d\u51c6\u786e\u7684\u751f\u7269\u5b66\u6ce8\u91ca\uff0c\u8fdb\u800c\u5f71\u54cd\u4e0b\u6e38\u5206\u6790\u7684\u53ef\u9760\u6027\u3002\u8fd9\u53ef\u628a\u5c0f\u4f19\u4f34\u4eec\u6101\u574f\u4e86\uff0c\u8d39\u5c3d\u4e5d\u725b\u4e8c\u864e\u4e4b\u529b\u5b8c\u6210\u4e86\u4ece\u964d\u7ef4\u3001\u805a\u7c7b\u5230\u6ce8\u91ca\uff0c\u7ed3\u679c\u5374\u4e0d\u77e5\u9053\u600e\u4e48\u89e3\u91ca\u4e86\u3002\u771f\u662f\u4e00\u987f\u64cd\u4f5c\u731b\u5982\u864e\uff0c\u4e00\u770b\u7ed3\u679c\u8981\u5fc3\u6897\uff01\u4e3a\u4e86\u514b\u670d\u8fd9\u4e9b\u5c40\u9650\u6027\uff0c\u5355\u7ec6\u80de\u5206\u6790\u5927\u4f6c\u56e2\u961f\u63d0\u51fa SIMBA\u3002\u5bf9\u4e8e\u6bcf\u4e2a\u4efb\u52a1\uff0cSIMBA \u6784\u9020\u4e00\u4e2a\u56fe\uff0c\u5176\u4e2d\u4e0d\u540c\u7684\u5b9e\u4f53(\u5373Cell\u548cFeature,\u5373\u57fa\u56e0\u6216\u8005Peaks)\u88ab\u8868\u793a\u4e3a\u8282\u70b9\uff0c\u8fd9\u4e9b\u5b9e\u4f53\u4e4b\u95f4\u7684\u5173\u7cfb\u88ab\u7f16\u7801\u4e3a\u8fb9(\u6784\u5efa\u4e00\u5f20\u56fe\uff09\u3002\u4e00\u65e6\u56fe\u5f62\u6784\u5efa\u5b8c\u6210\uff0cSIMBA \u7136\u540e\u5e94\u7528\u6e90\u81ea\u793e\u4ea4\u7f51\u7edc\u6280\u672f\u7684\u591a\u5173\u7cfb\u56fe\u5d4c\u5165\u7b97\u6cd5\u4ee5\u53ca\u57fa\u4e8e Softmax \u7684\u8f6c\u6362\u6765\u5c06\u8282\u70b9\u6216\u5b9e\u4f53\u5d4c\u5165\u5230\u516c\u5171\u7684\u4f4e\u7ef4\u7a7a\u95f4\u4e2d\uff0c\u5176\u4e2d\u7ec6\u80de\u548c\u7279\u5f81\u53ef\u4ee5\u6839\u636e\u5b83\u4eec\u7684\u8ddd\u79bb\u8fdb\u884c\u5206\u6790\u3002<\/p>\n\n\n\n

SIMBA\u53ef\u4ee5\u5728\u7edf\u4e00\u7684\u6846\u67b6\u4e0b\u89e3\u51b3\u5404\u79cd\u5355\u7ec6\u80de\u4efb\u52a1\uff0c\u5305\u62ec\uff1a\uff081\uff09\u964d\u7ef4;\uff082\uff09\u65e0\u805a\u7c7b\u6807\u8bb0\u68c0\u6d4b;\uff083\uff09\u591a\u6a21\u6001\u5206\u6790;\uff084\uff09\u6279\u6b21\u6821\u6b63\u548c\u7ec4\u5b66\u6574\u5408\u3002<\/p>\n\n\n\n

\u6709\u4ec0\u4e48\u751f\u4fe1\u5206\u6790\u4e0a\u7684\u95ee\u9898\u5927\u5bb6\u5c3d\u7ba1\u54a8\u8be2\u5927\u6d77\u54e5\uff01\u6ca1\u6709\u65f6\u95f4\u5b66\u4e60\u7684\u5c0f\u4f19\u4f34\u4eec\u4e5f\u4e0d\u8981\u7740\u6025\u54e6\uff01\u6709\u9700\u8981\u751f\u4fe1\u5206\u6790\u7684\u5c0f\u4f19\u4f34\u4eec\u4e5f\u53ef\u4ee5\u627e\u5927\u6d77\u54e5\u54e6\uff01\u7ec3\u4e86\u5341\u5e74\u751f\u4fe1\u5206\u6790\u7684\u5927\u6d77\u54e5\u5bf9\u4e8e\u751f\u4fe1\u5206\u6790\u77e5\u8bc6\u5df2\u7ecf\u5982\u9c7c\u5f97\u6c34\u4ece\u5206\u6790\u5230\u53ef\u89c6\u5316\u76f4\u5230\u4f60\u6ee1\u610f\u4e3a\u6b62\uff01<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

\u751f\u4fe1\u6570\u636e\u5904\u7406\u8d77\u6765\u5360\u7528\u5185\u5b58\u5b9e\u5728\u592a\u5927\u4e86\uff0c\u653e\u8fc7\u81ea\u5df1\u7684\u7535\u8111\u5427\uff01\u5927\u6d77\u54e5\u5728\u8fd9\u91cc\u7ed9\u5927\u5bb6\u9001\u4e0a\u798f\u5229\u4e86\uff0c\u6709\u9700\u8981\u670d\u52a1\u5668\u7684\u5c0f\u4f19\u4f34\u4eec\uff0c\u6b22\u8fce\u5927\u5bb6\u8054\u7cfb\u5927\u6d77\u54e5\uff0c\u4fdd\u8bc1\u670d\u52a1\u5668\u7684\u6027\u4ef7\u6bd4\u6700\u9ad8\u54e6<\/p>\n\n\n\n

\u4ee3\u7801\u6559\u7a0b<\/strong><\/strong><\/p>\n\n\n\n

\u5728\u672c\u6559\u7a0b\u4e2d\uff0c\u6211\u4eec\u5c06\u5229\u7528 10X PBMC scRNA-seq \u6570\u636e\u96c6\u6765\u5c55\u793a\u5728 scRNA-seq \u6570\u636e\u96c6\u4e0a\u8fd0\u884c SIMBA \u7684\u6b65\u9aa4\u3002\u8fd9\u91cc\u7684\u6570\u636e\u53ef\u4ee5\u66ff\u6362\u4e3a\u4f60\u81ea\u5df1\u9700\u8981\u5206\u6790\u7684\u6570\u636e\u96c6\uff0c\u53ea\u8981\u4fdd\u8bc1\u8f93\u5165\u6570\u636e\u7684\u683c\u5f0f\u4e3aanndata\u683c\u5f0f\u5373\u53ef\uff0c\u8fd9\u91cc\u6211\u4eec\u4f7f\u7528\u7684\u662fpbmc3k\u6570\u636e\u96c6\uff0c\u8fd9\u4e5f\u662f\u4e00\u4e2a\u975e\u5e38\u7ecf\u5178\u7684\u516c\u5171\u6570\u636e\u96c6\u3002\u8f93\u5165\u6570\u636e10\u591a\u4e2aG\uff0c\u4e0b\u8f7d\u94fe\u63a5\u79c1\u4fe1\u5927\u6d77\u54e5\u83b7\u53d6\u54e6\uff01<\/p>\n\n\n\n

1. \u5b89\u88c5SIMBA\u5305\u53ca\u5bfc\u5165\u76f8\u5173\u4f9d\u8d56\u5305<\/strong><\/strong><\/p>\n\n\n\n

\u5bf9\u4e8e\u9996\u6b21\u4f7f\u7528 conda \u7684\u7528\u6237\uff0c\u8bf7\u4f7f\u7528\u4ee5\u4e0b\u547d\u4ee4\u6267\u884c Bioconda \u7684\u4e00\u6b21\u6027\u8bbe\u7f6e\uff1a<\/p>\n\n\n\n

conda config –add channels defaults<\/p>\n\n\n\n

conda config –add channels bioconda<\/p>\n\n\n\n

conda config –add channels conda-forge<\/p>\n\n\n\n

conda config –set channel_priority strict<\/p>\n\n\n\n

\u8981\u4f7f\u7528 conda \u5b89\u88c5 simba\uff0c\u8bf7\u8fd0\u884c\uff1a<\/p>\n\n\n\n

conda install -c bioconda simba<\/p>\n\n\n\n

\u5efa\u8bae\uff1a\u5728\u65b0\u7684\u865a\u62df\u73af\u5883\u4e2d\u5b89\u88c5 simba\uff1a<\/p>\n\n\n\n

conda create -n env_simba simba<\/p>\n\n\n\n

conda activate env_simba<\/p>\n\n\n\n

\u8981\u5728 GitHub \u4e0a\u5b89\u88c5\u6700\u65b0\u7248\u672c\uff0c\u8bf7\u6267\u884c\u4ee5\u4e0b\u64cd\u4f5c\uff1a<\/p>\n\n\n\n

\u9996\u5148\u5b89\u88c5simba_pbg<\/p>\n\n\n\n

conda install -c bioconda simba_pbg<\/p>\n\n\n\n

\u7136\u540e\u8fd0\u884c\uff1a<\/p>\n\n\n\n

git clone https:\/\/github.com\/huidongchen\/simba.git<\/p>\n\n\n\n

pip install simba –user<\/p>\n\n\n\n

\u6216<\/p>\n\n\n\n

pip install git+https:\/\/github.com\/huidongchen\/simba<\/p>\n\n\n\n

import os<\/p>\n\n\n\n

import simba as si<\/p>\n\n\n\n

si.__version__<\/p>\n\n\n\n

workdir = ‘result_simba_rnaseq’<\/p>\n\n\n\n

si.settings.set_workdir(workdir)<\/p>\n\n\n\n

si.settings.set_figure_params(dpi=80,<\/p>\n\n\n\n

                              style=’white’,<\/p>\n\n\n\n

                              fig_size=[5,5],<\/p>\n\n\n\n

                              rc={‘image.cmap’: ‘viridis’})<\/p>\n\n\n\n

# make plots prettier<\/p>\n\n\n\n

from matplotlib_inline.backend_inline import set_matplotlib_formats<\/p>\n\n\n\n

set_matplotlib_formats(‘retina’)<\/p>\n\n\n\n

2. \u52a0\u8f7d\u793a\u4f8b\u6570\u636e\u5e76\u8fdb\u884c\u9884\u5904\u7406<\/strong><\/strong><\/p>\n\n\n\n

\u6211\u4eec\u8bfb\u53d6\u6d4b\u8bd5\u6570\u636e\uff0c\u968f\u540e\u6211\u4eec\u5bf9\u6d4b\u8bd5\u6570\u636e\u96c6\u8fdb\u884c\u8d28\u63a7\u548c\u8fc7\u6ee4\uff0c\u4ee5\u6ee4\u9664\u8868\u8fbe\u91cf\u8f83\u4f4e\u7684\u57fa\u56e0\uff0c\u8fd9\u6837\u53ef\u4ee5\u51cf\u5c0f\u6570\u636e\u7ef4\u5ea6\uff0c\u964d\u4f4e\u6570\u636e\u7a00\u758f\u6027\u5e76\u63d0\u9ad8\u4e0b\u6e38\u5206\u6790\u7684\u901f\u5ea6\u548c\u51c6\u786e\u6027\u3002<\/p>\n\n\n\n

adata_CG = si.datasets.rna_10xpmbc3k()<\/p>\n\n\n\n

adata_CG<\/p>\n\n\n\n

si.pp.filter_genes(adata_CG,min_n_cells=3)<\/p>\n\n\n\n

si.pp.cal_qc_rna(adata_CG)<\/p>\n\n\n\n

si.pl.violin(adata_CG,list_obs=[‘n_counts’,’n_genes’,’pct_mt’])<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

\u6211\u4eec\u6839\u636e\u4eba\u4e3a\u8bbe\u5b9a\u7684\u9608\u503c\u8fc7\u6ee4\u4f4e\u8d28\u91cf\u7ec6\u80de\uff0c\u5e76\u5bf9\u8ba1\u6570\u77e9\u9635\u7684\u8868\u8fbe\u503c\u505a\u6807\u51c6\u5316\u548c\u5bf9\u6570\u5316\u5904\u7406\uff0c\u7136\u540e\u8ba1\u7b97\u524d2000\u4e2a\u9ad8\u53d8\u57fa\u56e0\u5e76\u5bf9\u5176\u53ef\u89c6\u5316\u3002<\/p>\n\n\n\n

\u8bf7\u6ce8\u610f\uff0c\u662f\u5426\u8ba1\u7b97\u9ad8\u53d8\u57fa\u56e0\u5bf9\u56fe\u5d4c\u5165\u7684\u7ed3\u679c\u5e76\u6ca1\u6709\u672c\u8d28\u5f71\u54cd\uff0c\u4f46\u662f\u8ba1\u7b97\u9ad8\u53d8\u57fa\u56e0\u53ef\u4ee5\u964d\u4f4e\u6a21\u578b\u7684\u8ba1\u7b97\u65f6\u95f4\uff0c\u4e0e\u4e4b\u76f8\u5bf9\u4e8e\uff0c\u5982\u679c\u4ec5\u4f7f\u7528\u9ad8\u53d8\u57fa\u56e0\u8fdb\u884c\u56fe\u5d4c\u5165\u8ba1\u7b97\uff0c\u6700\u7ec8\u751f\u6210\u7684\u4f4e\u7ef4\u8868\u793a\u56fe\u4e2d\u4e5f\u53ea\u5305\u542b\u9ad8\u53d8\u57fa\u56e0\u3002<\/p>\n\n\n\n

si.pp.filter_cells_rna(adata,min_n_genes=100)<\/p>\n\n\n\n

si.pp.normalize(adata_CG,method=’lib_size’)<\/p>\n\n\n\n

si.pp.log_transform(adata_CG)<\/p>\n\n\n\n

\u4e5f\u53ef\u4ee5\u6267\u884c\u53ef\u53d8\u57fa\u56e0\u9009\u62e9\u6b65\u9aa4,\u4e5f\u53ef\u4ee5\u4e0d\u6267\u884c\uff0c\u4e0d\u5f71\u54cdR\u5305\u7684\u8ba1\u7b97\u7ed3\u679c\u3002<\/p>\n\n\n\n

si.pp.filter_cells_rna(adata,min_n_genes=100)<\/p>\n\n\n\n

si.pp.normalize(adata_CG,method=’lib_size’)<\/p>\n\n\n\n

si.pp.log_transform(adata_CG)<\/p>\n\n\n\n

3. \u79bb\u6563\u5316 RNA \u8868\u8fbe<\/strong><\/strong><\/p>\n\n\n\n

\u6b64\u6b65\u9aa4\u5c06\u975e\u96f6\u57fa\u56e0\u8868\u8fbe\u503c\u5206\u7ec4\u4e3a\u4e0d\u540c\u7684\u6c34\u5e73\uff0c\u540c\u65f6\u4fdd\u7559\u539f\u59cb\u5206\u5e03\uff0c\u4ece\u800c\u5728\u968f\u540e\u7684\u8bad\u7ec3\u4e2d\u8bc6\u522b\u4e0d\u540c\u6c34\u5e73\u7684\u57fa\u56e0\u8868\u8fbe\u3002\u79bb\u6563\u5316\u7684\u8868\u8fbe\u503c\u5c06\u7528\u4e8e\u6784\u5efa\u57fa\u56e0-\u7ec6\u80de\u56fe\u4e2d\u7684\u57fa\u56e0-\u7ec6\u80de\u8fb9\u7684\u6743\u91cd\u3002<\/p>\n\n\n\n

si.tl.discretize(adata_CG,n_bins=5)<\/p>\n\n\n\n

si.pl.discretize(adata_CG,kde=False)<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

4. \u751f\u6210\u7528\u4e8e\u8bad\u7ec3\u7684\u56fe\u5f62<\/strong><\/strong><\/p>\n\n\n\n

\u6bcf\u4e2a Anndata \u5bf9\u8c61\u4e2d\u7684\u89c2\u6d4b\u503c\u548c\u53d8\u91cf\u90fd\u8868\u793a\u4e3a\u8282\u70b9\u3002list_CG\u4e2d\u6307\u5b9a\u7684anndata\u5bf9\u8c61\u4e3a\u8f93\u5165\u7f16\u7801\u6570\u636e\uff0cadata.var_names\u548cdata.obs_names\u5c06\u88ab\u7f16\u7801\u4e3a\u8282\u70b9\uff0c\u5bf9\u4e8e\u57fa\u56e0\u8868\u8fbe\u6570\u636e\uff0c\u5728\u9ed8\u8ba4\u60c5\u51b5\u4e0b\uff0c\u4e0d\u540c\u6c34\u5e73\u7684\u57fa\u56e0\u8868\u8fbe\u662f\u7531\u4e0d\u540c\u6743\u91cd\u7684\u8282\u70b9-\u8282\u70b9\u4e4b\u95f4\u8fb9\u7f16\u7801\u7684\u3002<\/p>\n\n\n\n

si.tl.gen_graph(list_CG=[adata_CG],<\/p>\n\n\n\n

                layer=’simba’,<\/p>\n\n\n\n

                use_highly_variable=False,<\/p>\n\n\n\n

                dirname=’graph0′)<\/p>\n\n\n\n

5.\u6a21\u578b\u8bad\u7ec3<\/strong><\/strong><\/p>\n\n\n\n

\u63a5\u4e0b\u6765\u6211\u4eec\u53ef\u4ee5\u8bad\u7ec3\u6a21\u578b\u4e86\u3002\u4f5c\u8005\u7ed9\u51fa\u4e86\u4fee\u6539\u6a21\u578b\u8bad\u7ec3\u53c2\u6570\u7684\u4ee3\u7801\uff0c\u4f46\u662f\u5728\u4e00\u822c\u60c5\u51b5\u4e0b\uff0c\u6211\u4eec\u4f7f\u7528\u9ed8\u8ba4\u53c2\u6570\u5c31\u53ef\u4ee5\u5f88\u597d\u7684\u8bad\u7ec3\u6a21\u578b\u3002\u901a\u5e38\uff0c\u552f\u4e00\u53ef\u80fd\u9700\u8981\u8c03\u6574\u7684\u53c2\u6570\u662f\u6743\u91cd\u8870\u51cfwd\u3002\u5728\u5927\u591a\u6570\u60c5\u51b5\u4e0b\uff0c\u81ea\u52a8\u786e\u5b9a\u7684\u503c\uff08\u901a\u8fc7\u542f\u7528\u5b9e\u73b0\uff09\u53ef\u4ee5\u5145\u5206\u6267\u884c\uff1aauto_wd=True\u3002\u5982\u679c\u4e0d\u9700\u8981\u5bf9\u53c2\u6570\u8fdb\u884c\u8c03\u6574\uff0c\u5219\u53ea\u9700\u4f7f\u7528\u4ee5\u4e0b\u65b9\u6cd5\u5373\u53ef\u6267\u884c\u8bad\u7ec3\uff1a<\/p>\n\n\n\n

si.tl.pbg_train(auto_wd=True, save_wd=True, output=’model’)<\/p>\n\n\n\n

\u5728\u8fd9\u91cc\uff0c\u6211\u4eec\u6f14\u793a\u4e86\u5982\u4f55\u5728\u5fc5\u8981\u65f6\u4fee\u6539\u4e0e\u8bad\u7ec3\u76f8\u5173\u7684\u53c2\u6570\u3002<\/p>\n\n\n\n

# \u66f4\u6539\u53c2\u6570<\/p>\n\n\n\n

dict_config = si.settings.pbg_params.copy()<\/p>\n\n\n\n

# dict_config[‘wd’] = 0.015521<\/p>\n\n\n\n

dict_config[‘wd_interval’] = 10 # we usually set `wd_interval` to 10 for scRNA-seq datasets for a slower but finer training<\/p>\n\n\n\n

dict_config[‘workers’] = 12 #The number of CPUs.<\/p>\n\n\n\n

## \u5f00\u59cb\u8bad\u7ec3<\/p>\n\n\n\n

si.tl.pbg_train(pbg_params = dict_config, auto_wd=True, save_wd=True, output=’model’)<\/p>\n\n\n\n

\u6211\u4eec\u4e5f\u53ef\u4ee5\u53ef\u89c6\u5316\u51fa\u6a21\u578b\u8bad\u7ec3\u6307\u6807\u7684\u53d8\u5316\u60c5\u51b5\u3002\u901a\u8fc7\u7ed8\u5236\u8bad\u7ec3\u6307\u6807\u4ee5\u76d1\u63a7\u8fc7\u5ea6\u62df\u5408\u5e76\u8bc4\u4f30\u6700\u7ec8\u6a21\u578b\u3002\u7406\u60f3\u60c5\u51b5\u4e0b\uff0c\u7ecf\u8fc7\u4e00\u5b9a\u6570\u91cf\u7684\u8bad\u7ec3\u540e\uff0c\u5ea6\u91cf\u66f2\u7ebf\u5e94\u8be5\u4fdd\u6301\u7a33\u5b9a\u3002\u9ed8\u8ba4\u60c5\u51b5\u4e0b\uff0c\u8fd9\u884c\u4ee3\u7801\u4f1a\u663e\u793a\u4e86\u8bad\u7ec3\u635f\u5931(\u8d8a\u4f4e\u8d8a\u597d)\u4ee5\u53ca\u9a8c\u8bc1\u635f\u5931(\u8d8a\u4f4e\u8d8a\u597d)\u548c MRR (\u8d8a\u9ad8\u8d8a\u597d)\u3002<\/p>\n\n\n\n

si.pl.pbg_metrics(fig_ncol=1)<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

\u5982\u679c\u9700\u8981\u8fdb\u4e00\u6b65\u7684\u53c2\u6570\u8c03\u6574\uff0c\u53ef\u4ee5\u901a\u8fc7\u66f4\u65b0\u53d8\u91cf\u6765\u5b9e\u73b0\u5728\u540c\u4e00\u56fe\u4e0a\u8bad\u7ec3\u65f6\u4f7f\u7528\u66f4\u65b0\u7684\u53c2\u6570\u751f\u6210\u65b0\u7684\u5d4c\u5165\uff0c\u4f8b\u5982\uff0coutput<\/p>\n\n\n\n

si.tl.pbg_train(pbg_params = dict_config, auto_wd=True, save_wd=True, output=”model2″)<\/p>\n\n\n\n

\u5982\u679c\u9700\u8981\uff0c\u4e00\u65e6\u8bad\u7ec3\u53c2\u6570\u6700\u7ec8\u786e\u5b9a\uff0c\u53ef\u4ee5\u4f7f\u7528\u4ee5\u4e0b\u51fd\u6570\u66f4\u65b0 PBG \u8bad\u7ec3\u53c2\u6570\uff1a<\/p>\n\n\n\n

si.settings.set_pbg_params(dict_config)<\/p>\n\n\n\n

\u8981\u52a0\u8f7d\u8bad\u7ec3\u7684\u7ed3\u679c\uff0c\u53ef\u4ee5\u6267\u884c\u4ee5\u4e0b\u6b65\u9aa4\uff1a\u9ed8\u8ba4\u60c5\u51b5\u4e0b\uff0c\u5b83\u4f7f\u7528\u5b58\u50a8\u5728.setting.pbg_params<\/p>\n\n\n\n

# load back ‘graph0’<\/p>\n\n\n\n

si.load_graph_stats()<\/p>\n\n\n\n

# load back ‘model’ that was trained on ‘graph0’<\/p>\n\n\n\n

si.load_pbg_config()<\/p>\n\n\n\n

\u7528\u6237\u8fd8\u53ef\u4ee5\u6307\u5b9a\u4e0d\u540c\u7684\u8def\u5f84\uff0c\u4f8b\u5982\uff0c<\/p>\n\n\n\n

# load back ‘graph1’<\/p>\n\n\n\n

si.load_graph_stats(path=’.\/result_simba_rnaseq\/pbg\/graph1\/’)<\/p>\n\n\n\n

# load back ‘model2’ that was trained on ‘graph1’<\/p>\n\n\n\n

si.load_pbg_config(path=’.\/result_simba_rnaseq\/pbg\/graph1\/model2\/’)<\/p>\n\n\n\n

6. \u8bad\u7ec3\u540e\u5206\u6790<\/strong><\/strong><\/p>\n\n\n\n

# \u8bfb\u53d6\u4ece\u4ecePBG\u57f9\u8bad\u83b7\u5f97\u7684\u5b9e\u4f53\u5d4c\u5165\u3002<\/p>\n\n\n\n

dict_adata = si.read_embedding()<\/p>\n\n\n\n

dict_adata<\/p>\n\n\n\n

adata_C = dict_adata[‘C’]  #\u7ec6\u80de\u5d4c\u5165<\/p>\n\n\n\n

adata_G = dict_adata[‘G’]  # \u57fa\u56e0\u5d4c\u5165<\/p>\n\n\n\n

7. \u53ef\u89c6\u5316\u7ec6\u80de\u548c\u57fa\u56e0\u7684\u5d4c\u5165<\/strong><\/strong><\/p>\n\n\n\n

\u63a5\u4e0b\u6765\u6211\u4eec\u9996\u5148\u53ef\u89c6\u5316\u51fa\u7ec6\u80de\u7684Umap\u56fe\uff0c\u57fa\u4e8e\u8fd950\u7ef4\u4f4e\u7ef4\u8868\u793a\uff0c\u6211\u4eec\u53ef\u4ee5\u5c06\u5176\u4f5c\u4e3aumap\u8ba1\u7b97\u7684\u8f93\u5165\uff0c\u5e76\u4e14\u5f97\u5230\u5176Umap\u5750\u6807\uff0c\u5e76\u53ef\u89c6\u5316\u3002<\/p>\n\n\n\n

##\u6dfb\u52a0\u7ec6\u80de\u7c7b\u578b\u6ce8\u91ca<\/p>\n\n\n\n

adata_C.obs[‘celltype’] = adata_CG[adata_C.obs_names,:].obs[‘celltype’].copy()<\/p>\n\n\n\n

adata_<\/p>\n\n\n\n

palette_celltype={‘B’:’#1f77b4′,<\/p>\n\n\n\n

                  ‘CD4 T’:’#ff7f0e’,<\/p>\n\n\n\n

                  ‘CD8 T’:’#279e68′,<\/p>\n\n\n\n

                  ‘Dendritic’:”#aa40fc”,<\/p>\n\n\n\n

                  ‘CD14 Monocytes’:’#d62728′,<\/p>\n\n\n\n

                  ‘FCGR3A Monocytes’:’#b5bd61′,<\/p>\n\n\n\n

                  ‘Megakaryocytes’:’#e377c2′,<\/p>\n\n\n\n

                  ‘NK’:’#8c564b’}<\/p>\n\n\n\n

si.tl.umap(adata_C,n_neighbors=15,n_components=2)<\/p>\n\n\n\n

si.pl.umap(adata_C,color=[‘celltype’],<\/p>\n\n\n\n

           dict_palette={‘celltype’: palette_celltype},<\/p>\n\n\n\n

           fig_size=(6,4),<\/p>\n\n\n\n

           drawing_order=’random’)<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

\u7136\u540e\u6211\u4eec\u53ef\u4ee5\u8ba1\u7b97\u5e76\u5f97\u5230\u7ec6\u80de\u548c\u57fa\u56e0\u7684\u8054\u5408\u5d4c\u5165\uff0c\u5e76\u5c06\u5176\u53ef\u89c6\u5316\u5728Umap\u56fe\u4e0a\u3002<\/p>\n\n\n\n

# \u5c06\u7ec6\u80de\u548c\u57fa\u56e0\u5d4c\u5165\u540c\u4e00\u7a7a\u95f4<\/p>\n\n\n\n

adata_all = si.tl.embed(adata_ref=adata_C,list_adata_query=[adata_G])<\/p>\n\n\n\n

adata_all.obs.head()<\/p>\n\n\n\n

## \u6dfb\u52a0\u7ec6\u80de\u548c\u57fa\u56e0\u6ce8\u91ca<\/p>\n\n\n\n

adata_all.obs[‘entity_anno’] = “”<\/p>\n\n\n\n

adata_all.obs.loc[adata_C.obs_names, ‘entity_anno’] = adata_all.obs.loc[adata_C.obs_names, ‘celltype’]<\/p>\n\n\n\n

adata_all.obs.loc[adata_G.obs_names, ‘entity_anno’] = ‘gene’<\/p>\n\n\n\n

adata_all.obs.head()<\/p>\n\n\n\n

palette_entity_anno = palette_celltype.copy()<\/p>\n\n\n\n

palette_entity_anno[‘gene’] = “#607e95”<\/p>\n\n\n\n

si.tl.umap(adata_all,n_neighbors=15,n_components=2)<\/p>\n\n\n\n

si.pl.umap(adata_all,color=[‘id_dataset’,’entity_anno’],<\/p>\n\n\n\n

           dict_palette={‘entity_anno’: palette_entity_anno},<\/p>\n\n\n\n

           drawing_order=’original’,<\/p>\n\n\n\n

           fig_size=(6,5))<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

\u4f5c\u8005\u540c\u6837\u63d0\u4f9b\u4e86\u5de5\u5177\uff0c\u5e2e\u52a9\u6211\u4eec\u53ef\u89c6\u5316\u51fa\u4e0d\u540c\u57fa\u56e0\u5728Umap\u56fe\u4e0a\u7684\u4f4d\u7f6e\uff0c\u5e76\u4e14\u7ed9\u51fa\u6587\u5b57\u8868\u793a\u3002<\/p>\n\n\n\n

\u56fe\u4e2d\u9664\u4e86GAPDH\u57fa\u56e0\u548cB2M\u57fa\u56e0\u662f\u7ba1\u5bb6\u57fa\u56e0\uff0c\u5176\u4ed6\u57fa\u56e0\u90fd\u662f\u7279\u5b9a\u7ec6\u80de\u4e9a\u7fa4\u7684\u6807\u8bb0\u57fa\u56e0\u3002\u5728Umap\u56fe\u4e0a\uff0c\u5b9e\u9645\u4e0a\u8fd9\u4e9bMarker\u57fa\u56e0\u7684Umap\u5750\u6807\u90fd\u5904\u4e8e\u5bf9\u5e94\u7684\u4e9a\u7fa4\u5185\uff0c\u800c\u7ba1\u5bb6\u57fa\u56e0\u7684\u5750\u6807\u5219\u4e0d\u5177\u6709\u4e9a\u7fa4\u7279\u5f02\u6027\u3002<\/p>\n\n\n\n

#\u5728\u5171\u5d4c\u5165\u7a7a\u95f4\u4e2d\u7a81\u51fa\u4e00\u4e9b\u6807\u8bb0\u57fa\u56e0<\/p>\n\n\n\n

marker_genes = [‘IL7R’, ‘CD79A’, ‘MS4A1’, ‘CD8A’, ‘CD8B’, ‘LYZ’, ‘CD14’,<\/p>\n\n\n\n

                ‘LGALS3’, ‘S100A8’, ‘GNLY’, ‘NKG7’, ‘KLRB1’,<\/p>\n\n\n\n

                ‘FCGR3A’, ‘MS4A7’, ‘FCER1A’, ‘CST3’, ‘PPBP’]<\/p>\n\n\n\n

si.pl.umap(adata_all[::-1,],color=[‘entity_anno’],dict_palette={‘entity_anno’: palette_entity_anno},<\/p>\n\n\n\n

           drawing_order=’original’,<\/p>\n\n\n\n

           texts=marker_genes + [‘GAPDH’, ‘B2M’],<\/p>\n\n\n\n

           show_texts=True,<\/p>\n\n\n\n

           fig_size=(8,6))<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

\u6211\u4eec\u8fd8\u53ef\u4ee5\u6309\u7167\u4ee5\u4e0b\u6b65\u9aa4\u4ec5\u53ef\u89c6\u5316\u90e8\u5206\u57fa\u56e0\uff08\u4f8b\u5982\uff0c\u53ef\u53d8\u57fa\u56e0\uff09\u800c\u4e0d\u662f\u6240\u6709\u57fa\u56e0\uff1a<\/p>\n\n\n\n

# \u83b7\u53d6\u53ef\u53d8\u57fa\u56e0<\/p>\n\n\n\n

si.pp.select_variable_genes(adata_CG, n_top_genes=3000)<\/p>\n\n\n\n

var_genes = adata_CG.var_names[adata_CG.var[‘highly_variable’]].tolist()<\/p>\n\n\n\n

# \u83b7\u5f97\u7ec6\u80de\u548c\u53ef\u53d8\u57fa\u56e0\u7684SIMBA\u5305\u57cb<\/p>\n\n\n\n

adata_all2 = adata_all[list(adata_C.obs_names) + var_genes,].copy()<\/p>\n\n\n\n

# \u4f7f\u7528UMAP\u53ef\u89c6\u5316<\/p>\n\n\n\n

si.tl.umap(adata_all2,n_neighbors=15,n_components=2)<\/p>\n\n\n\n

si.pl.umap(adata_all2,color=[‘entity_anno’],<\/p>\n\n\n\n

           dict_palette={‘entity_anno’: palette_entity_anno},<\/p>\n\n\n\n

           drawing_order=’original’,<\/p>\n\n\n\n

           fig_size=(6,5))<\/p>\n\n\n\n

\u8ba1\u7b97SIMBA\u6307\u6807\uff0c\u5305\u62ec\u6700\u5927\u503c\u3001\u57fa\u5c3c\u6307\u6570\u3001\u6807\u51c6\u5dee\u548c\u71b5\uff0c\u4ee5\u8bc4\u4f30\u57fa\u56e0\u7684\u7ec6\u80de\u7c7b\u578b\u7279\u5f02\u6027\uff1amax\uff1a\u503c\u8d8a\u9ad8\u8868\u793a\u7ec6\u80de\u7c7b\u578b\u7279\u5f02\u6027\u8d8a\u9ad8<\/p>\n\n\n\n

\u57fa\u5c3c\u7cfb\u6570\uff1a\u503c\u8d8a\u9ad8\u8868\u793a\u7ec6\u80de\u7c7b\u578b\u7279\u5f02\u6027\u8d8a\u9ad8<\/p>\n\n\n\n

STD\uff1a\u503c\u8d8a\u9ad8\u8868\u793a\u7ec6\u80de\u7c7b\u578b\u7279\u5f02\u6027\u8d8a\u9ad8<\/p>\n\n\n\n

\u71b5\uff1a\u503c\u8d8a\u4f4e\u8868\u793a\u7ec6\u80de\u7c7b\u578b\u7279\u5f02\u6027\u8d8a\u9ad8<\/p>\n\n\n\n

adata_cmp = si.tl.compare_entities(adata_ref=adata_C, adata_query=adata_G)<\/p>\n\n\n\n

adata_cmp<\/p>\n\n\n\n

# \u57fa\u56e0\u53ef\u4ee5\u6839\u636e\u4e00\u4e2a\u6216\u591a\u4e2a\u6307\u6807\u8fdb\u884c\u6392\u540d<\/p>\n\n\n\n

adata_cmp.var.head()<\/p>\n\n\n\n

# SIMBA\u6307\u6807\u53ef\u4ee5\u4f7f\u7528\u4ee5\u4e0b\u51fd\u6570\u8fdb\u884c\u53ef\u89c6\u5316:<\/p>\n\n\n\n

si.pl.entity_metrics(adata_cmp,<\/p>\n\n\n\n

                     x=’max’,<\/p>\n\n\n\n

                     y=’gini’,<\/p>\n\n\n\n

                     show_contour=False,<\/p>\n\n\n\n

                     texts=marker_genes + [‘GAPDH’, ‘B2M’],<\/p>\n\n\n\n

                     show_texts=True,<\/p>\n\n\n\n

                     show_cutoff=True,<\/p>\n\n\n\n

                     size=5,<\/p>\n\n\n\n

                     text_expand=(1.3,1.5),<\/p>\n\n\n\n

                     cutoff_x=1.,<\/p>\n\n\n\n

                     cutoff_y=0.3,<\/p>\n\n\n\n

                     save_fig=False)<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

\u4f7f\u7528SIMBA \u6761\u5f62\u7801\u56fe\u57fa\u4e8e\u8fb9\u7f18\u7f6e\u4fe1\u5ea6\u53ef\u89c6\u5316\u7279\u5f81\u7ec6\u80de\u6982\u7387:-\u4e0d\u5e73\u8861\u8868\u793a\u7ec6\u80de\u7c7b\u578b\u7279\u5f02\u6027\u57fa\u56e0-\u5747\u5300\u5206\u5e03\u8868\u793a\u975e\u7279\u5f02\u6027\u57fa\u56e0\u3002<\/p>\n\n\n\n

# \u6dfb\u52a0\u7ec6\u80de\u6ce8\u91ca<\/p>\n\n\n\n

adata_cmp.obs[‘celltype’] = adata_CG.obs.loc[adata_cmp.obs_names,’celltype’]<\/p>\n\n\n\n

list_genes = [‘CST3’, ‘NKG7’, ‘MS4A1’, ‘GAPDH’]<\/p>\n\n\n\n

si.pl.entity_barcode(adata_cmp,<\/p>\n\n\n\n

                     layer=’softmax’,<\/p>\n\n\n\n

                     entities=list_genes,<\/p>\n\n\n\n

                     anno_ref=’celltype’,<\/p>\n\n\n\n

                     show_cutoff=True,<\/p>\n\n\n\n

                     cutoff=0.001,<\/p>\n\n\n\n

                     palette=palette_celltype,<\/p>\n\n\n\n

                     fig_size=(6, 2.5),<\/p>\n\n\n\n

                     save_fig=False)<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

\u540c\u6837\u7684\u57fa\u56e0\u5217\u8868\u4e5f\u53ef\u4ee5\u5728UMAP\u4e0a\u53ef\u89c6\u5316\uff0c\u4ee5\u786e\u8ba4\u5b83\u4eec\u7684\u7ec6\u80de\u7c7b\u578b\u7279\u5f02\u6027<\/p>\n\n\n\n

adata_CG.obsm[‘X_umap’] = adata_C[adata_CG.obs_names,].obsm[‘X_umap’].copy()<\/p>\n\n\n\n

si.pl.umap(adata_CG,<\/p>\n\n\n\n

           color=[‘CST3’, ‘NKG7’, ‘MS4A1’, ‘GAPDH’],<\/p>\n\n\n\n

           drawing_order=’sorted’,<\/p>\n\n\n\n

           size=5,<\/p>\n\n\n\n

           alpha=0.9,<\/p>\n\n\n\n

           fig_ncol=4,<\/p>\n\n\n\n

           fig_size=(4,4),<\/p>\n\n\n\n

           save_fig=False)<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

\u5c0f\u7ed3<\/strong><\/strong><\/p>\n\n\n\n

SIMBA\u7684\u4f18\u70b9\u662f\u53ef\u4ee5\u6574\u5408\u4e0d\u540c\u65b9\u6cd5\u5f97\u5230\u7684\u57fa\u56e0\uff0c\u6bd4\u5982\u7ec6\u80de\u901a\u8baf\u3001MOFA\uff08\u591a\u7ec4\u5b66\u56e0\u5b50\u5206\u6790\uff0c\u4e00\u79cd\u591a\u7ec4\u5b66\u5206\u6790\u7b97\u6cd5\uff09\u7684feature\u3001GLUE\u7684GRN\uff08\u9ad8\u6b4c\u8001\u5e08\u5f00\u53d1\u7684scRNA-seq+scATAC-seq\u914d\u5bf9\u7b97\u6cd5\u751f\u6210\u7684\u8f6c\u5f55\u56e0\u5b50\u8c03\u63a7\u7f51\u7edc\uff09\u3001pySCENIC\u751f\u6210\u7684\u8c03\u8282\u5b50\u6d3b\u6027\u77e9\u9635\u3001\u8fd8\u6709\u7ec6\u80de\u7684marker\uff0c\u53ef\u4ee5\u5728embedding\u5c42\u9762\uff0c\u7814\u7a76\u4ed6\u4eec\u7684\u76f8\u5173\u6027\u548c\u7279\u5f81\u3002\u6b64\u5916\uff0cSIMBA\u5728scATAC-seq\u5206\u6790\uff0cscRNA+scATAC-seq\u6574\u5408\u5206\u6790\u7684\u5c42\u9762\uff0c\u53ef\u80fd\u4f1a\u6709\u66f4\u597d\u7684\u6548\u679c\u3002\u6700\u540e\u5927\u6d77\u54e5\u7ed9\u5927\u5bb6\u4ecb\u7ecd\u4e00\u4e2a\u4e91\u5de5\u5177\uff01\u540c\u5b66\u4eec\u5982\u679c\u89c9\u5f97\u81ea\u5df1\u7684\u4ee3\u7801\u6c34\u5e73\u4e00\u822c\uff0c\u5bf9\u4e8e\u5f88\u591a\u7684\u53c2\u6570\u4e0d\u77e5\u9053\u600e\u4e48\u6539\uff0c\u53ef\u4ee5\u4f53\u9a8c\u4e00\u4e0b\u6211\u4eec\u7684\u4e91\u751f\u4fe1\u5c0f\u5de5\u5177\uff0c\u53ea\u9700\u8f93\u5165\u6570\u636e\uff0c\u5373\u53ef\u8f7b\u677e\u751f\u6210\u6240\u9700\u56fe\u8868\uff0c\u5b57\u4f53\u5927\u5c0f\u3001\u6807\u9898\u7b49\u4e5f\u53ef\u4e00\u952e\u66f4\u6539\u3002\u611f\u5174\u8da3\u7684\u5c0f\u4f19\u4f34\u53bb\u4e91\u751f\u4fe1\uff08http:\/\/www.biocloudservice.com\/home.html\uff09\u4f53\u9a8c\u4e00\u4e0b\u5427\uff01<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

<\/p>\n","protected":false},"excerpt":{"rendered":"

\u524d\u8a00 \u8c08\u5230\u5355\u7ec6\u80de\u5206\u6790\uff0c\u5c11\u4e0d\u4e86\u7684\u5c31\u662f\u805a\u7c7b\uff01\u805a\u7c7b\uff01\u5728\u805a\u7c7b\uff01\uff01\u4e0d\u5e78\u7684\u662f\uff0c\u8fd9\u4e9b\u201c\u4ee5\u805a\u7c7b\u4e2d\u5fc3\u201d\u5206\u6790\u65b9\u6cd5\u4f9d\u8d56\u4e8e\u51c6\u786e\u5b9a\u4e49\u7684\u805a […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"jetpack_post_was_ever_published":false,"_jetpack_newsletter_access":""},"categories":[1],"tags":[],"jetpack_featured_media_url":"","_links":{"self":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/60891"}],"collection":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=60891"}],"version-history":[{"count":1,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/60891\/revisions"}],"predecessor-version":[{"id":60903,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/60891\/revisions\/60903"}],"wp:attachment":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=60891"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=60891"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=60891"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}