pylimma.convest

pylimma.convest(p, niter=100, plot=False, report=False, tol=1e-06, file='')[source]

Estimate pi0 using a convex decreasing density estimate.

Port of R limma’s convest. The file argument controls where the diagnostic report is written when report=True: the default empty string writes to stdout (matching R’s file="" convention); any non-empty string is opened as a path.

Parameters:

  • p (array_like) – Observed p-values in [0, 1].

  • niter (int, default 100) – Number of iterations of the EM-like algorithm.

  • plot (bool, default False) – If True, draw the estimated density at each iteration.

  • report (bool, default False) – If True, print a per-iteration diagnostic table.

  • tol (float, default 1e-6) – Accuracy of the bisection search for the convex combination.

  • file (str, default "") – Destination for the report. Empty string -> stdout.

Returns:

Estimated proportion of true null hypotheses.

Return type:

float