RegioneR library in R
1
0
Entering edit mode
2.2 years ago

Hello!

I try to use regioneR library in R for permutation test. I get result, where z-score is equal to 79 or -90 and p-value is equal to 0.002 for both z-scores. Number of permutations is 500. I don't understand how those z-score and p-value are calculated.

Thanks!

R permutation test • 766 views
ADD COMMENT
0
Entering edit mode

Yes, and there are no explanation why for different z-score there are the same p-value.

ADD REPLY
0
Entering edit mode

They are not the same z-score.

ADD REPLY
0
Entering edit mode
2.2 years ago
ATpoint 62k

These are bootstraping p-values. The second formula on this page here gives a reference where this comes from.

From the source code of the regioneR package (permTest function) they implement it as:

pval <- (sum(orig.ev >= rand.ev, na.rm = TRUE) + 1) / (num.valid.values + 1)


So the p-value, here testing that the overlap (or whatever you test) is greater for the observed events compared to the randomly-permutated events, is the number of times that the observed event is larger or equal divided by the number of permutations.

The Z-score is the deviation of the observed events by the mean of the permutated events (corrected for standard deviation) so basically a significant event should have a large Z-score since the observed event should be quite different from the mean of the random events.

zscore <- round((orig.ev - mean(rand.ev, na.rm = TRUE))/stats::sd(rand.ev,  na.rm = TRUE), 4)

ADD COMMENT

Login before adding your answer.

Traffic: 1037 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6