Preprocessing Clustering

Source: src/single_cell_python_tools/preprocessing/_clustering.py

These helpers are exported through sctl.pp.

leiden_clustering

sctl.pp.leiden_clustering(
    adata,
    number_of_neighbors=10,
    number_of_PC=40,
    leiden_res=1,
    key_added="leiden",
    **parameters,
)

Build a neighbor graph, run UMAP, and calculate Leiden clusters.

rename_leiden_clusters

sctl.pp.rename_leiden_clusters(
    adata,
    rename_cluster=False,
    new_cluster_names=None,
    new_obs_key="Cell_Clusters",
    make_new_obs_key_categorical=True,
    reorder_cluster_names=False,
    new_cluster_names_order=None,
    **parameters,
)

Map Leiden cluster IDs to provided cluster names and optionally reorder the resulting categorical labels.

Legacy Rename Helpers

• rename_leiden_clusters_old_old(...)

• rename_leiden_clusters_old(...)

These legacy helpers remain importable from the current source.

leiden_cluster_sil_score

sctl.pp.leiden_cluster_sil_score(
    adata,
    leiden_res=1,
    n_jobs=8,
    savefig=False,
    output_dir="./figures/",
    output_prefix="adata",
    **parameters,
)

Calculate silhouette scores for a Leiden resolution and optionally save the resulting figure.

silhouette_walk_Largest_drop

sctl.pp.silhouette_walk_Largest_drop(
    adata,
    leiden_res_start=0.025,
    leiden_res_end=2,
    leiden_res_step=0.025,
    print_per_step=False,
    n_jobs=8,
    savetable=False,
    savefig=False,
    output_dir="./figures/",
    output_prefix="adata",
)

Scan Leiden resolutions and report the largest silhouette-score drop.

silhouette_walk_4_Largest_drops

sctl.pp.silhouette_walk_4_Largest_drops(
    adata,
    leiden_res_start=0.025,
    leiden_res_end=2,
    leiden_res_step=0.025,
    print_per_step=False,
    n_jobs=8,
    savetable=False,
    savefig=False,
    output_dir="./figures/",
    output_prefix="adata",
)

Scan Leiden resolutions and report the four largest silhouette-score drops.