[BioC] sensitivity of maxT correction

[A S]

Hi All,
I have a question regarding resampling-based corrections (maxT or minP) that 
are part of multtest.
I have searched past archives and found some info By Prof Dudoit on this 
thread
https://stat.ethz.ch/pipermail/bioconductor/2003-July/001947.html
But I still need a clarification.
It appears that when the resampling null distribution of t statistics vary 
across tests, maxT or minP single-step procedures will not be equally 
sensitive to all tests and this can be overcome by step-down
minp/maxT. Does this mean maxT step-down procedure will be identically 
sensitive to all tests?
Or is it that step-down method offers higher sensitivity over single-step by 
adjusting the unequal
distributions across the tests, but the sensitivity across tests would still 
  vary though to a smaller extent .
Thank you for your response in advance.

Archana

_________________________________________________________________
One and only Ash. Find out all about her. Only on MSN Search