The number of overlaps are more than the true set

[KopalliV]

Hi, Thank you very much for regioneR.
Just a question,
I am trying to check the significance of overlapping variants between my variant set and a true set.
I have around 37k variants in my true set but I get more than 44k observed overlaps, even if I use the function count.once.
This is the code I used:

gn <- read.csv("Chr.len",sep="\t", header=F)
names(gn)<-c("chr","start","end")
bt <- read.csv("true_set.pos",sep="\t", header=F)
names(bt)<-c("chr","start","end")
at <- read.csv("variants.pos",sep="\t", header=F)
names(at)<-c("chr","start","end")
pt <- permTest(A=at, B=bt, randomize.function=randomizeRegions, evaluate.function=numOverlaps, ntimes=100, genome = gn, count.once=TRUE)
plot(pt)

'at' has 179656 variant positions and bt which is the true set has 37555 variant positions
pt[["numOverlaps"]][["observed"]]
[1] 43896
If my B has 37555 variants, then shouldn't my observed overlap count be <= 37555? Am I missing something basic here?