| RPS-package(RPS) | R Documentation |
Reproducibility probability score incorporates measurement variability across labs for gene selection. A gene with a large RPS means that it is more likely to be differentially expressed; and if other labs were to perform the same experiment,the differential expression of this gene is more likely to be reaffirmed.
| Package: | RPS |
| Type: | Package |
| Version: | 1.0 |
| Date: | 2006-05-21 |
| License: | GPL |
Guixian Lin<glin2@uiuc.edu> and Sheng Zhong<szhong@ad.uiuc.edu>
RPS.load, RPS.readnew, RPS.select
load("ABI.RData") #load precomputed correlations of genes for different platforms using MAQC microarray dataset.Here the platform is "ABI"
#Right now,one of the following platform(ABI,AFFY,AGL,GEH or ILM) can be specified.
##After loading "ABI.RData",two R objects( corr and probenames) are in the current enviroment.
##samples information
test.sample.info=c(rep(1,5),rep(2,5)) #two samples,first 5 replicates are from sample A, the next 5 replicates are from sample B.
K=10 # the number of labs to be simulated,recommendly between 30 and 50.
n.replicate=5 #the replicates for each sample
p.value=1e-03 # p-value threshold for computing RPS
FC=2.5 # fold change threshold for computing RPS
MAQC.test=RPS.readnew("ABI_12091.txt",transformed=F) #read the actural microarray data to compute RPS;Here The data set include 3 labs,4 samples , and 5 replicates
#if transform==F, then the actural data need log2-tranformation.
results=RPS.select(MAQC.test[,2:11],test.sample.info,corr,n.replicate,test="t",p.value,FC,K,flag=T) #MAQC.test[,2:11] are samples A and B with 5 replications repectively from Lab 1,
#and results is a list,which includes RPS and K simulation data
##If the test data set only contains partly genes in MAQC data set or you are only interested in some genes,then you can compute
## the RPS for those genes only
## For example, only first 200 genes in your test data set, use :
## subindex= is.element(probenames,MAQC.test[1:200,1])
## subcorr=list(corr.Labs=corr$corr.Labs[subindex],var.matrix=corr$var.matrix[,subindex])
##results=RPS.select(MAQC.test[1:200,2:11],test.sample.info,subcorr,n.replicate,test="t",p.value,FC,K,flag=T)
summary(results$RPS)
#or choose to show RPS for each probe set
probeSetName=MAQC.test[,1] ##First column is the probe set names
data.frame(probeSetName,results$RPS)[1:20,] ##Here is an example to show RPS for the first 20 probe set
#obtain the summary expression data for the j-th new imaginary lab
j=2;#make sure j<=K
simu.dataOneLab=results$simu.data[,1:(n.replicate*2)+(j-1)*(n.replicate*2)]
colnames( simu.dataOneLab)<-c(rep("s1",n.replicate),rep("s2",n.replicate)) #first n.replicates are sample A,and the next n.replicates are sample B
summary(simu.dataOneLab)