BGmix fits a variety of Bayesian hierarchical models for finding differential gene expression between 2 or more experimental conditions.
BGmix uses a C++ routine to fit the chosen model via an MCMC algorithm. Files are written to a sub-directory in the working directory. The package includes R functions for reading the results into R, and several plotting functions and functions for estimating error rates.
Lewin, A., Bochkina, N. and Richardson, S. (2007), Fully Bayesian mixture model for differential gene expression: simulations and model checks. http://www.bgx.org.uk/publications.html
## Note this is a very short MCMC run! ## For good analysis need proper burn-in period. data(ybar,ss) outdir <- BGmix(ybar, ss, c(8,8), nburn=0, niter=100, nthin=1,trace.pred=1) ## Basic plot of parameters params <- ccParams(outdir) plotBasic(params,ybar,ss) ## plots of FDR and related quantities fdr <- calcFDR(params) par(mfrow=c(1,2)) plotFDR(fdr) ## plots of Bayesian p-values ## for predictive checks of mixture prior pred <- ccPred(outdir,q.trace=TRUE) plotPredChecks(pred$pval.ybar.mix2,params$pc,probz=0.5) ## plots of predictive density superimposed on data plotMixDensity(params,pred,ybar,ss)