preprocess.apply_mask

preprocess.apply_mask(data, *, platform=None, kind='general')

Set masked probes to NaN. Not called anywhere automatically.

.. warning::

Every clock in this registry was fitted on unmasked data, before most of these masks existed. Removing a probe at score time does not recover a truer number: it deletes an input the coefficients expect and hands the gap to the imputer. Run :func:mask_report first and decide with the coefficient-mass column in front of you.

Masking is the right default for an EWAS and a real decision for a clock.