Closed zx8754 closed 5 years ago
Here is a better example, where all tests give "significant" pvalue, while MB is zero:
#skatO
Range RANGE N_INFORMATIVE NumVar NumPolyVar Q rho Pvalue
geneX x x 11 11 58761.7 1 0.000204181
#MadsonBrowning
Range RANGE N_INFORMATIVE NumVar NumPolyVar NumPerm ActualPerm Stat NumGreater NumEqual PermPvalue
geneX x x 11 11 10000 10000 16.6566 0 0 0
#CMC
Range RANGE N_INFORMATIVE NumVar NumPolyVar NonRefSite Pvalue
geneX x x 11 11 22 0.000121977
#Zeggini
Range RANGE N_INFORMATIVE NumVar NumPolyVar Pvalue
geneX x x 11 11 0.000204544
The permutation p-value is exactly zero, and it is calculated based on 10,000 permutations. I think if the number of permutation increases, you will get non-zero p-values for MadsonBrowning test.
I see, this answers my question then, zero means value is very close to zero and is significant. Thank you.
MadsonBrowning test results have zero values for "PermPvalue", does this mean Pvalue is so small that it is rounded to 0 and we should consider it as "significant", or this Pvalue is supposed to be
NA
?Here is subset of example output, see gene2: