Python implementation of of the irreproducible discovery rate (short IDR) proposed in [LI2011].
IDR was originally published as R package idr.
The local IDR and the proportion of the reproducibility component can be done as follows:
import pyidr idr = pyidr.IDR() idr.fit(dataset1, dataset2) local_idr = idr.predict(dataset1,dataset2) proportion_of_reproducibility = idr.proportion
The correspondence profile used for diagnostics can be caluclated and plotted with the :class:`CorrespondenceProfile` class as follows:
import pyidr profile = pyidr.CorrespondenceProfile(rankx, ranky, t) profile.plot_diagnostics()
[LI2011] | Li, Qunhua, et al. "Measuring reproducibility of high-throughput experiments." The annals of applied statistics 5.3 (2011): 1752-1779. |