I have two questions about DMRcate:
1) Why the smoothed Y statistic (SKY) is far elevated from limma Y statistic? extracted the SKY from the code.
2) From the code, the parameter "betacutoff" has never been used. Is it?
1) Yes that's right - this is normal for kernel aggregation given it's a local sum. Remember that SKY is moderated by SK and S_KK using Satterthwaite, and also the FDR is indexed against the number of DMLs - both control for this inflation.
2) Thanks for spotting - that filter must have been squelched a few updated ago. I've put it back in: https://github.com/timpeters82/DMRcate-devel/blob/master/R/dmrcate.R#L176-L180.
Thanks for the explanation.
Login before adding your answer.
Use of this site constitutes acceptance of our User Agreement and Privacy