baySeq-package {baySeq} | R Documentation |
This package is intended to identify differential expression in high-throughput 'count' data, such as that derived from next-generation sequencing machines. We achieve this by empirical bayesian methods, first bootstrapping to estimate prior parameters from the data and then assessing posterior likelihoods of the models proposed.
Package: | baySeq |
Type: | Package |
Version: | 1.1.1 |
Date: | 2009-16-05 |
License: | GPL-3 |
LazyLoad: | yes |
To use the package, construct a countData
object and use
the functions documented in getPriors to empirically determine
priors on the data. Then use the functions documented in
getLikelihoods to establish posterior likelihoods for the models
proposed. A few convenience functions, getTPs
and
topCounts
are also included.
The package (optionally) makes use of the 'snow' package for parallelisation of computationally intensive functions. This is highly recommended for large data sets.
See the vignette for more details.
Thomas J. Hardcastle
Maintainer: Thomas J. Hardcastle <tjh48@cam.ac.uk>
Hardcastle T.J., and Kelly, K. baySeq: Empirical Bayesian Methods For Identifying Differential Expression In Sequence Count Data. BMC Bioinformatics (2010)
# See vignette for more examples. # load test data data(simData) # replicate structure of data replicates <- c("simA", "simA", "simA", "simA", "simA", "simB", "simB", "simB", "simB", "simB") # define hypotheses on data groups <- list(NDE = c(1,1,1,1,1,1,1,1,1,1), DE = c(1,1,1,1,1,2,2,2,2,2)) # construct 'countData' object CD <- new("countData", data = simData, replicates = replicates, groups = groups) #estimate library sizes for countData object libsizes(CD) <- getLibsizes(CD) # estimate prior distributions on 'countData' object using negative binomial # method. Other methods are available - see getPriors CDPriors <- getPriors.NB(CD, cl = NULL) # estimate posterior likelihoods for each row of data belonging to each hypothesis CDPost <- getLikelihoods(CDPriors, cl = NULL) # display the rows of data showing greatest association with the second # hypothesis (differential expression) topCounts(CDPost, group = "DE", number = 10) # find true positive selection rate getTPs(CDPost, group = "DE", TPs = 1:100)[1:100]