effectSize_bin

oasis_stat.effectSize_bin(X, c, f)[source]

Computes the effect size measure from the OASIS paper. Binarizes samples into positive and negative groups based on the sign of the column embedding.

Parameters:

X (numpy.ndarray) – The input count matrix, IxJ.
c (numpy.ndarray) – The column embedding vector, J-dimensional.
f (numpy.ndarray) – The row embedding vector, I-dimensional.

Returns:

The two-group effect size measure.

Return type:

float

Description

Computes the effect size, based on OASIS’s identified f and c embeddings. This quantity is bounded between 0 and 1, and provides a measure of how stark the difference is between the row distributions of the two groups (partitioned by the sign of c).