CRISGI Class Initialization

Function

The CRISGI class initializes and preprocesses gene expression data for downstream analysis, including background network construction and feature selection. It provides methods for data preprocessing, background network loading, and supports various interaction inference methods.

Parameters

Name	Type	Description
adata	AnnData	Annotated data matrix (cells × genes) to be analyzed.
bg_net	array or None	Optional. Precomputed background network matrix.
bg_net_score_cutoff	int	Score threshold for filtering background network edges. Default is 850.
genes	list or None	Optional. List of gene names to include in the analysis.
n_hvg	int or None	Number of highly variable genes to select. Default is 5000.
n_pcs	int	Number of principal components for dimensionality reduction. Default is 30.
interactions	array or None	Optional. Predefined gene-gene interactions to use for background network construction.
n_threads	int	Number of threads to use for computation. Default is 5.
interaction_methods	list	List of methods for inferring gene interactions. Default: ['prod'].
organism	str	Organism name (e.g., 'human'). Default is 'human'.
class_type	str	Type of classification task (e.g., 'time'). Default is 'time'.
dataset	str	Dataset identifier. Default is 'test'.
out_dir	str	Output directory for results. Default is './out'.

Return Type

CRISGI object

Returns

Initializes a CRISGI object with preprocessed data and background network ready for downstream analysis.

Attributes Set

• adata: Preprocessed AnnData object.
• interaction_methods: List of interaction inference methods.
• organism: Organism name.
• n_threads: Number of computation threads.
• dataset: Dataset identifier.
• class_type: Classification type.
• out_dir: Output directory path.
• bg_net_score_cutoff: Background network score cutoff.

Example

import anndata as ad
from crisgi import CRISGI

# Load your single-cell data into an AnnData object
adata = ad.read_h5ad('example_data.h5ad')

# Initialize CRISGI with default parameters
crisgi = CRISGI(
    adata=adata,
    n_hvg=3000,
    n_pcs=20,
    organism='human',
    class_type='time',
    dataset='my_dataset',
    out_dir='./crisgi_output'
)

# The object is now ready for further analysis
print(crisgi.adata)
print(crisgi.interaction_methods)