TailPP {BGmix} | R Documentation |
For differential expression models with unstructured priors (no mixture prior), calculates tail posterior probabality and FDR, and plots a histogram. Uses whole posterior distributions of likelihood parameters (found by 'ccTrace') and posterior means of hyperparameters (found by 'ccParams').
TailPP(res, nreps, params, paired=F, alpha=0.05, N = 5000, prec=0.05, p.cut = 0.7, plots = T, plot.pi0=F)
res |
list object output from 'ccTrace' |
nreps |
vector length 2 containing the number of replicates in each condition |
params |
list object output from 'ccParams' |
paired |
logical. TRUE for paired design, FALSE for unpaired. |
alpha |
parameter of the tail posterior probability (1-alpha/2 quantile) |
N |
simulation size for tail posterior probability under H0 |
prec |
parameter used when estimating CDF of tail posterior probability under H0 |
p.cut |
calculate FDR only for cutoffs on tail posterior probability > p.cut |
plots |
logical. if TRUE, makes plots of the histogram of tail posterior probability with the null density and of FDR |
plot.pi0 |
logical. if TRUE, diagnostic plot of the estimated pi0 at different locations and the median estimate |
tpp |
vector of tail posterior probabilities with parameter alpha, one per gene |
FDR |
(smoothed) estimate of FDR for all (distinct) cutoffs > p.cut |
pi0 |
estimated proportion of observations under the null |
pp0 |
simulations under the null |
Natalia Bochkina
Bochkina N., Richardson S. (2007) Tail posterior probability for inference in pairwise and multiclass gene expression data. Biometrics. http://www.blackwell-synergy.com/doi/abs/10.1111/j.1541-0420.2006.00807.x
FDRplotTailPP
,histTailPP
,EstimatePi0
data(ybar, ss) nreps <- c(8,8) ## Note this is a very short MCMC run! ## For good analysis need proper burn-in period. outdir <- BGmix(ybar, ss, nreps, jstar=-1, nburn=0, niter=100, nthin=1) params <- ccParams(outdir) res <- ccTrace(outdir) tpp.res <- TailPP(res, nreps, params, plots = FALSE) histTailPP(tpp.res) FDRplotTailPP(tpp.res, plot.TP = TRUE)