Open Robaina opened 1 year ago
maxT doesn't work because there are very small pathways that are very likely to have a high representation in the shuffled cluster, thus the maxT distribution contains very large numbers and the corrected p-value is always very large. Perhaps try minP procedure instead?
See:
https://stats.stackexchange.com/questions/166510/controlling-fwer-using-minp-maxt-methods, report both permutation pvalue and corrected permutaton p-value
https://mne.tools/stable/generated/mne.stats.permutation_t_test.html