xchrom.tl.calc_auc_pr

xchrom.tl.calc_auc_pr(true_matrix: array, pred_matrix: array, mtype: Literal['overall', 'percell', 'perpeak'] = 'overall') → dict[source]

Calculate the AUROC and AUPRC metrics for given true and predicted matrix. Shape of true_matrix and pred_matrix must be consistent, cells x peaks.

Parameters:

  • true_matrix (np.array) – The true label matrix (cells x peaks), 0/1 matrix, 0 for no signal, 1 for peak signal.

  • pred_matrix (np.array,) – The predicted probability matrix (cells x peaks), 0-1 matrix for peak signal.

  • mtype (Literal['overall', 'percell', 'perpeak'] = 'overall') – The type of metric to calculate, optional ‘overall’, ‘percell’, ‘perpeak’

Examples

>>> auc, prc = calc_auc_pr(true_matrix, pred_matrix, mtype='overall')
>>> print(f"The AUROC is {auc:.4f}, The AUPRC is {prc:.4f}")