bumphunter P values and significance testing
0
0
Entering edit mode
cardenasca ▴ 40
@cardenasca-9929
Last seen 4.3 years ago

Hi,

I'm interested in the significance testing performed by bumphunter. Namely, as of now bumphunter outputs two different pvalues: 1) p.value along with the fwer and 2) p.valueArea along with the fwerArea.

My understanding is that these are derived from bootstraping set at B (i.e. B=10000 for my case).

The issues is that for the region we are finding there is no agreement between the fwer or the fwer Area using a cut off of P<0.05 for the fwer: See below:

> head(dmr.pick.tops)
chr     start       end       value      area cluster indexStart indexEnd fwerArea
chr1 205818956 205819492 -0.06205978 0.3723587   14664      33501    33506 6
L clusterL      p.value      fwer     p.valueArea       fwerArea
6 9           5.884939e-06   0.021    0.0001457223        0.41

Any advice on how to proceed or what significance value to report?

Thanks!

minfi bumphunter • 956 views