Qunyuan Zhang writes.... > Hi, > > We just finished an initial inverstigation (50000-gene Affymetrix, 15 > cancered people and 10 normal people). 40 genes' RNA expressional levels > were found significantly different between the two groups (by two sample > t tests, p values corrected). We are now planning a second-stage > experiment to validate this finding. We want to do power analysis and > sample size calculation, especially want to know how many peoples should > be included in the second-stage experiment. > > Besides the function Power.t.test(), is there any other functions in > any packages availabe in bioConductor for this kind of experimantal > design problems? It depends how you intend to summarize your data. Will you be calculating some type of summary score based on the expression of these 40 genes? If so, you can easily perform a power calculation based on the expected means and standard deviations as estimated by your initial experiment. Or will you test each of the 40 genes independently? In that case, you will need to take into account some type of multiple comparisons adjustment (e.g. bonferroni, etc.). > > Thanks, > > Qunyuan Zhang > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor This email message, including any attachments, is for the so...{{dropped}}

On 1/26/06 1:20 PM, "Kort, Eric" <eric.kort at="" vai.org=""> wrote: > > > Qunyuan Zhang writes.... >> Hi, >> >> We just finished an initial inverstigation (50000-gene Affymetrix, 15 >> cancered people and 10 normal people). 40 genes' RNA expressional > levels >> were found significantly different between the two groups (by two > sample >> t tests, p values corrected). We are now planning a second-stage >> experiment to validate this finding. We want to do power analysis and >> sample size calculation, especially want to know how many peoples > should >> be included in the second-stage experiment. >> >> Besides the function Power.t.test(), is there any other functions in >> any packages availabe in bioConductor for this kind of experimantal >> design problems? > > It depends how you intend to summarize your data. Will you be > calculating some type of summary score based on the expression of these > 40 genes? If so, you can easily perform a power calculation based on > the expected means and standard deviations as estimated by your initial > experiment. Or will you test each of the 40 genes independently? In > that case, you will need to take into account some type of multiple > comparisons adjustment (e.g. bonferroni, etc.). Just to add a bit, with only 40 genes, it may be more cost-effective to use PCR rather than another set of arrays. Sean