Added the cluster module

- Added the `cluster` and `filtered` module for clustering the cells

omicverse/utils/__init__.py

@@ -35,3 +35,4 @@
 from ._heatmap import *
 from ._roe import roe
 from ._paga import cal_paga,plot_paga
+from ._cluster import cluster

omicverse/utils/_cluster.py

@@ -0,0 +1,53 @@
+#聚类
+from sklearn.mixture import GaussianMixture
+import scanpy as sc
+import pandas as pd
+import anndata
+
+#初始化聚类位置，这个很重要
+def get_initial_means(X, n_components,init_params, r):
+    # Run a GaussianMixture with max_iter=0 to output the initialization means
+    gmm = GaussianMixture(
+        n_components=n_components, init_params=init_params, tol=1e-9, max_iter=0, random_state=r
+    ).fit(X)
+    return gmm.means_
+
+
+def cluster(adata:anndata.AnnData,method:str='leiden',
+            use_rep:str='X_pca',random_state:int=1024,
+            n_components=None, **kwargs):
+
+    if method=='leiden':
+        sc.tl.leiden(adata,**kwargs)
+    elif method=='louvain':
+        sc.tl.louvain(adata,**kwargs)
+    elif method=='GMM':
+        if n_components is None:
+            print('You need to input the `n_components` when methods is `GMM`')
+            return
+        print(f"""running GaussianMixture clustering""")
+        data=adata.obsm[use_rep].copy()
+        ini = get_initial_means(data,n_components, 'k-means++', 0)
+        gmm = GaussianMixture(n_components = n_components,random_state=random_state,
+                     means_init=ini, **kwargs)
+        gmm.fit(data)
+        adata.obs['gmm_cluster']=gmm.predict(data)
+        adata.obs['gmm_cluster']=adata.obs['gmm_cluster'].astype(str)
+
+        #new_num=adata.obs['gmm_cluster'].value_counts()[adata.obs['gmm_cluster'].value_counts()>10].shape[0]
+        #adata.obs.loc[adata.obs['gmm_cluster'].isin(adata.obs['gmm_cluster'].value_counts()[adata.obs['gmm_cluster'].value_counts()<10].index.tolist()),'gmm_cluster']='-1'
+
+        #adata.obs['gmm_cluster']=adata.obs['gmm_cluster'].astype('category')
+        #adata.obs['gmm_cluster'].cat.categories=pd.Index(list(range(len(adata.obs['gmm_cluster'].cat.categories))))
+
+        print(f"""finished: found {n_components} clusters and added
+    'gmm_cluster', the cluster labels (adata.obs, categorical)""")
+
+def filtered(adata:anndata.AnnData,
+             cluster_key:str,
+             cluster_minsize:int=10):
+    new_num=adata.obs[cluster_key].value_counts()[adata.obs[cluster_key].value_counts()<cluster_minsize].shape[0]
+    adata.obs.loc[adata.obs[cluster_key].isin(adata.obs[cluster_key].value_counts()[adata.obs[cluster_key].value_counts()<cluster_minsize].index.tolist()),cluster_key]='-1'
+    adata.obs[cluster_key]=adata.obs[cluster_key].astype('category')
+    print(f"""filtered {new_num} clusters and changed the cluster labels to '-1'(adata.obs, categorical)""")
+

omicverse_guide/docs/Release_notes.md

@@ -246,3 +246,4 @@
 
 ### utils Module:
 - Introduced `plot_pca_variance_ratio` for visualizing the ratio of PCA variance.
+- Added the `cluster` and `filtered` module for clustering the cells