| initialize.priors {RPA} | R Documentation |
Creates a template for prior parameters using non-informative (or user-specified) priors. After creating the template, probe-specific priors can be set by modifying the default values. Current version supports priors are only for the probe-specific variances (i.e. priors alpha, beta).
initialize.priors(abatch, sets, alpha = 1e-6, beta = 1e-6, d = NULL)
abatch |
An AffyBatch object. |
sets |
Vector listing the probesets for which the prior template is created. |
alpha |
Default template values for the alpha prior |
beta |
Default template values for the beta prior |
d |
Not used, included for later compatibility. |
Probesets that do not have predefined priors are analyzed using noninformative alpha, beta (1e-6). See the source code for RPA.iteration for details.
An instance of 'rpa.priors' class.
Leo Lahti <leo.lahti@tkk.fi>
Probabilistic Analysis of Probe Reliability in Differential Gene Expression Studies with Short Oligonucleotide Arrays. Lahti et al., TCBB/IEEE, to appear. See http://www.cis.hut.fi/projects/mi/software/RPA/
require(affy)
require(affydata)
data(Dilution)
# List probesets to investigate
sets = geneNames(Dilution)[1:5]
# Create a template for prior parameters using default priors
my.priors = initialize.priors(Dilution, sets, alpha = 1e-6, beta = 1e-6)
# Modify the template to provide user-specified prior for one of the
# probes
# high value implies an unreliable probe
set = sets[[1]]
probe.idx = 5
my.priors[[set]]$beta[[probe.idx]] = 10
my.priors[[set]]$alpha[[probe.idx]] = 10
# Run RPA using the predefined priors
# NOTE: priors are only used with sigma2.method = "basic"
rpa.results <- RPA.pointestimate(Dilution,sets,
priors=my.priors,
sigma2.method = "basic",
d.method = "basic")
# This toy example shows the probe reliability values for the probeset
# where the user-specified prior was set for one of the probes
barplot(rpa.results$sigma2[[set]])