We have been rerunning some QC and found that samples that previously passed on detection p value filtering are now all failing. Closer inspection we worked out that the detection p values are identical to the beta values.
Trawling through the code I can see that this error is introduced in NChannelSetToMethyLumiSet2 function. You can see below that the calculation of the detection p-values duplicates the code for the calculation of the beta values.
https://github.com/schalkwyk/wateRmelon/blob/0e85b0d606e22da5ad613bf83b0f79389f6e7f6b/R/readEPIC.R#L609C11-L611C61
betas = methylated/(methylated + unmethylated)
# should this not be +100 in denom?
pvals = methylated/(methylated + unmethylated)
We have been rerunning some QC and found that samples that previously passed on detection p value filtering are now all failing. Closer inspection we worked out that the detection p values are identical to the beta values.
Trawling through the code I can see that this error is introduced in
NChannelSetToMethyLumiSet2function. You can see below that the calculation of the detection p-values duplicates the code for the calculation of the beta values.https://github.com/schalkwyk/wateRmelon/blob/0e85b0d606e22da5ad613bf83b0f79389f6e7f6b/R/readEPIC.R#L609C11-L611C61