sincei.TopicModels module#

class sincei.TopicModels.TOPICMODEL(adata, n_topics, binarize=False, smart_code='lfu', n_passes=1, n_workers=1)[source]#

Bases: object

Computes LSA or LDA for a given matrix and returns the cell-topic matrix.

Parameters#

adataAnnData: AnnData object containing the data matrix in adata.X, with cells in adata.obs_names and regions in adata.var_names.
n_topicsint: Number of Topics / Principal Components for modeling.
binarizebool, optional: If True, the input matrix will be binarized (default is False). Recommended for LDA.
smart_codestr: SMART (System for the Mechanical Analysis and Retrieval of Text) code for weighting of input matrix for TFIDF. Only valid for the LSA model. The default ("lfu") corresponds to "log"TF * IDF, and "pivoted unique" normalization of document length. For more information, see: https://en.wikipedia.org/wiki/SMART_Information_Retrieval_System
n_passesint, optional: Number of passes for the LDA model. Default is 1.
n_workersint, optional: Number of workers for the LDA model. Default is 1.

runLSA()[source]#: Computes LSA for a given matrix and updates the TOPICMODEL object.

runLDA()[source]#: Computes LDA model for a given matrix and updates the TOPICMODEL object.

get_cell_topic(pop_sparse_cells=False)[source]#

Get cell-topic matrix from the TOPICMODEL object.