scanpy.tl

API: Tools

sc.tl.pca()

此函数使用方法同 sc.pp.pca()，此函数本身已废弃

https://scanpy.readthedocs.io/en/stable/generated/scanpy.pp.pca.html#scanpy.pp.pca

PCA, linear dimensional reduction

sc.pp.pca(data=, svd_solver="arpack")

• 1. n_comps=: number of PC to be caculated
• 2. layer=: expression matrix used; default=None, .X will be used
• 3. svd_solver=: SVD solver to be used

结果存放在 .obsm 中，使用 .obsm['X_pca'] 来提取

sc.tl.tsne()

https://scanpy.readthedocs.io/en/stable/generated/scanpy.tl.tsne.html#scanpy.tl.tsne

tSNE

sc.tl.tsne(adata, n_pcs=, random_state=, use_rep=)

• 1. n_pcs=: number of top PCs to use; default=None, all genes will be used
• 2. random_state=: 设置随机种子数，以保证每次结果的一致性
• 3. use_rep=: key in .obsm be used; default=None, .obsm.X_pca will be used; after data integration, .obsm.X_pca_harmony should be assigned

结果存放在 .obsm 中，使用 .obsm['X_tsne'] 来提取

sc.tl.umap()

https://scanpy.readthedocs.io/en/stable/generated/scanpy.tl.umap.html

UMAP

sc.tl.umap(adata, min_dist=, spread=, random_state=)

• 1. min_dist=: 最终降维结果的紧密程度；数值越小，越紧密，有利于显示局部结果；数值越大越松散，有利于显示整体结果；default=0.5; 取值介于 [0, 1] 之间
• 2. spread=: 与 min_dist= 结合使用；控制最终点的散布程度；数值越小，点越集中；default=1
• 3. random_state=: 设置随机种子数，以保证每次结果的一致性

识别细胞亚群：sc.tl.umap(adata, min_dist=0.1, spread=1.0, random_state=149)
整体结构分析：sc.tl.umap(adata, min_dist=0.8, spread=2.0, random_state=149)

结果存放在 .obsm 中，使用 .obsm['X_umap'] 来提取

sc.tl.leiden()

https://scanpy.readthedocs.io/en/stable/generated/scanpy.tl.leiden.html#scanpy.tl.leiden

Cluster cells using the Leiden algorithm

sc.tl.leiden(adata, resolution=, random_state=)

• 1. resolution=: 用来控制分辨率；default=1
• 2. random_state=: 设置随机种子数，以保证每次结果的一致性

sc.tl.leiden(adata, resolution=1, random_state=149)

sc.tl.rank_genes_groups()

Expects logarithmized data.
https://scanpy.readthedocs.io/en/stable/generated/scanpy.tl.rank_genes_groups.html#scanpy.tl.rank_genes_groups

call the DEGs of every cluster:

sc.tl.rank_genes_groups(scanpy_object, groupby=, method=, n_genes=, use_raw=, layer=)

• 1. groupby=: 分组依据；按照 clustering 结果分组则是 'leiden'
• 2. method=: 用来进行计算的方法；default=t-tset
• 3. n_genes=: number of gene to return; degault=None, need to be assigned
• 4. use_raw=: use the .raw.X for caculation; default=None, .raw.X will be used
• 5. layer=: key of .layer to be used for caculated

sc.tl.rank_genes_groups(scanpy_object, groupby='leiden', method='t-test', n_genes=20, use_raw=True, layer=None)

计算数据存放在 .uns 中

数据提取并将格式转换为 DataFrame:

>>> kk=pd.DataFrame(scanpy_object.uns['rank_genes_groups']['names'])     '#提取基因名
>>> display(kk.shape)
>>> kk.head()     #示例一，Gene names

>>> pp=pd.DataFrame(scanpy_object.uns['rank_genes_groups']['pvals'])     #提取p-values
>>> display(pp.shape)
>>> pp.head()     #示例二，P-values

结果中的 logfoldchanges 为 log2()

Gene names

P-values

Differential expression analysis between two celltypes:

sc.tl.rank_genes_groups(scanpy_object, groupby='column_name', groups=['celltype_name_1'], reference='celltype_name_2', method='wilcoxon', layer='log1p')

NOTE: 结果是 groups= vs. reference=

提取差异表达后的结果:

deg_results = sc.get.rank_genes_groups_df(scanpy_oject, group="celltype_name_1")

NOTE: 结果是 groups= vs. reference=