Suppose when I write the results of a microarray experiment fit using limma to file. In doing this I decide to adjust the p-values using the BH method for adjusting the p-values. As I understand it, this method assesses the significance of a result assuming some prior rate of false discovery (type 1 error). I read the 1995 paper for the BH method but I am still confused on what exactly determines this rate? Is it done in a parametric way or non-paramterically? Joseph Cornish Post-Bac IRTA NIAID EVPS [[alternative HTML version deleted]]

The "adjusted p-value" for the i_th smallest p-value is mp/i if this is less than 1 and greater than the "adjusted p-value" for the (i-1) smallest p-value. It is truncated at 1.0 from above and at the value for the (i-1) value below (so that it always increases as p increases). This is the estimated FDR for the list of genes you obtain if you reject all tests with p less than or equal the i_th smallest p-value. --Naomi At 12:51 PM 7/3/2013, Cornish, Joseph (NIH/NIAID) [F] wrote: >Suppose when I write the results of a microarray experiment fit >using limma to file. In doing this I decide to adjust the p-values >using the BH method for adjusting the p-values. As I understand it, >this method assesses the significance of a result assuming some >prior rate of false discovery (type 1 error). I read the 1995 paper >for the BH method but I am still confused on what exactly determines >this rate? Is it done in a parametric way or non-paramterically? > >Joseph Cornish >Post-Bac IRTA >NIAID EVPS > > > [[alternative HTML version deleted]] > >_______________________________________________ >Bioconductor mailing list >Bioconductor at r-project.org >https://stat.ethz.ch/mailman/listinfo/bioconductor >Search the archives: >http://news.gmane.org/gmane.science.biology.informatics.conductor