thor.analy.SPARKX

class thor.analy.SPARKX(rscript_path='R/run_SPARKX.R', **kwargs)

Bases: object

Class for running SPARKX.

Parameters:

rscript_path (str, default: "R/run_SPARKX.R") – Path to the R script for running SPARKX.

Methods

RUN_SPARKX_R	Run SPARKX.
compute_pattern_mean	Compute the mean expression of each gene module.
hierarchy_clustering	Run hierarchical clustering with sklearn's AgglomerativeClustering on the residual matrix.
kmeans_clustering	Run k-means clustering with sklearn's KMeans on the residual matrix.
load_gene_modules	Load the gene modules of SPARKX.
load_result	Load the result of SPARKX.

RUN_SPARKX_R(adata_path=None, layer=None, out_path=None)

Run SPARKX.

Parameters:

• adata_path (str) – Path to the AnnData object.

• layer (str, default: None) – Layer of the AnnData object to use.

• out_path (str, default: None) – Path to the output directory.

static compute_pattern_mean(adata, data, pattern, obskey_prefix)

Compute the mean expression of each gene module.

Parameters:

• adata (AnnData) – (n_sig_genes x n_cells)

• data (dataframe (n_sig_genes x n_cells)) –

• pattern (dataframe (n_sig_genes x 1), column is cluster, index is gene) –

hierarchy_clustering(**hc_kwargs)

Run hierarchical clustering with sklearn’s AgglomerativeClustering on the residual matrix.

Parameters:

hc_kwargs (dict) – Keyword arguments for AgglomerativeClustering.

Returns:

labels – Cluster labels.

Return type:

array (n_cells,)

kmeans_clustering(n_patterns, **kmeans_kwargs)

Run k-means clustering with sklearn’s KMeans on the residual matrix.

Parameters:

• n_patterns (int) – Number of clusters.

• kmeans_kwargs (dict) – Keyword arguments for KMeans.

Returns:

labels – Cluster labels.

Return type:

array (n_cells,)

load_gene_modules(pattern_prefix='SP')

Load the gene modules of SPARKX.

Parameters:

pattern_prefix (str, default: "SP") – Prefix of the gene modules.

load_result()

Load the result of SPARKX.

Returns:

residual – (n_genes x n_cells)

Return type:

dataframe