pyCellPhenoX utilities¶

pyCellPhenoX.utils.balanced_sample module¶

pyCellPhenoX.utils.balanced_sample.balanced_sample(group, subset_percentage)¶

Perform balanced sampling on a DataFrame group.

Parameters:

group (DataFrame) – The DataFrame or group to sample from.
subset_percentage (float) – The fraction of the group to sample (between 0.0 and 1.0).

Returns:

A randomly sampled fraction of the group, based on the given percentage.

Return type:

DataFrame

pyCellPhenoX.utils.check_indices.check_indices(a, b)¶

Check that the indices are matching for the dataframes and assign indices if they aren’t.

Parameters:

Return type:

a, b (pd.DataFrame, pd.DataFrame)

pyCellPhenoX.utils.reducedim.reduceDim(reducMethod, reducMethodParams, expression_mat)¶

Call the reduction method specified by user

Parameters:

Returns:

one matrix if PCA selected, tuple of matrices if NMF selected

Return type:

matrix/matrices

pyCellPhenoX.utils.select_num_components.select_number_of_components(eigenvalues, var)¶

Find the number of the components based on the percentage of accumulated variance

Parameters:

eigenvalues (array) – array of eigenvalues (explained variances) for the components
var (float) – desired proportion of variance explained

Returns:

number of components

Return type:

int

pyCellPhenoX.utils.select_optimal_k.select_optimal_k(X, min_k, max_k)¶

Select optimal k (number of components) and generate elbow plot for silhouette score

Parameters:

X (dataframe) – the marker by cell matrix to be decomposed
numberOfComponents (int) – number of components or ranks to learn (if -1, then we will select k)
min_k (int) – alternatively, provide the minimum number of ranks to test
max_k (int) – and the maximum number of ranks to test

Returns:

optimal k for decomposition

Return type:

int