i am a PhD student in Bioinformatics and i'm dealing with some statistics issue in my reasearch; i was hoping you could lend me a hand

I am dealing with a pool of datapoints (let's call it Q).

I selected a subset of Q according to some particular characteristics, i will call this dataset Qs.

because the datapoints in Q do not follow any particular distribution, i wanted to generate a background distribution by randomly sampling Q. i would obtain several randomly generated datasets (Qr) of the same size as Qs and compare each time my Qs with the randomly sampled Qr in order to obtain the probability that Qs might have occurred at random when sampling Q.

In order to perform the comparison between the two datasets (Qs and Qr), i use a wilcoxon test with a threshold of 0.01. this however puts me in an awkward spot.

from my multiple sampling trials i can basically determine the probability that my Qs will "fail" the wilcoxon test with a threshold of 0.01 when compared to a randomly sampled Qr.

is there a way to consolidate these 2 p-values into one? can i treat them as probabilities and multiply them to obtain the overall probability of error?

Thank you very much in advance

Tito