prot2D (version 1.10.0)

Sim.Prot.2D: Simulate 2D Gel volume data

Description

This function simulates realistic 2D Gel volume data, based on parameters estimate from real dataset.

Usage

Sim.Prot.2D(data, nsp=nrow(data), nr=10, p0=0.1, s2_0=0.2 ,d0=3)

Arguments

data

either a dataframe or a matrix of 2D Gel Volume data. Usually the output of Norm.qt.

nsp

an integer. Number of spots (i.e. rows) to generate (should be equal or less than the number of rows of data).

nr

an integer. Number of replicates for each conditions.

p0

proportion of spots to be differentially expressed (defaults to 0.1).

s2_0

estimator of the standard deviation of the spots distribution (defaults to 0.2).

d0

degree of freedom of the Chi-squared distribution used in the calculation of the distribution (defaults to 3).

Value

returns an ExpressionSet of simulated volume data (log2 transformed) with 2 conditions ("Cond1" and "Cond2" in phenoData) slot of the ExpressionSet.The spots differentially generated can be retrieve with notes.

Details

Volume data are computed following these steps (see Smyth, 2004 and Artigaud et al, 2013 for details) :

1. Log2 mean volumes from data are computed for each spot.
2. Means are used as input parameters in order to simulate a normal distribution (with no differential expression between conditions) for each spots with standard deviations, computed as described by Smyth (2004).
3. p0 of the spots are randomly picked for introducing differential expression in both conditions (p0/2 in each condition).

If nsp is less than the number of rows in data, the nsp first columns of data are used as input parameters.

References

• Artigaud, S., Gauthier, O. & Pichereau, V. (2013) "Identifying differentially expressed proteins in two-dimensional electrophoresis experiments: inputs from transcriptomics statistical tools." Bioinformatics, vol.29 (21): 2729-2734.
• Smyth, G.K. (2004) "Linear models and empirical bayes methods for assessing differential expression in microarray experiments." Statistical Applications in Genetics and Molecular Biology, vol. 3: Article 3.

See Also

variousT.Prot,

Examples

#Simulate data based on "pecten"
data(pecten)
data(pecten.fac)

pecten.norm <- Norm.qt(pecten, n1=6, n2=6, plot=FALSE)
Sim.data <- Sim.Prot.2D(data=pecten.norm, nsp=700, nr=10, p0=0.1, s2_0=0.2, d0=3)

#Compare different methods for finding diffenrentially expressed proteins
res.stud <- ttest.Prot(Sim.data, fdr.thr=0.1, plot=FALSE)
res.mo <- modT.Prot(Sim.data, fdr.thr=0.1, plot=FALSE)

a <- featureNames(res.stud) 
a # Names of the spots selected for a student's t-test with an FDR of 0.1
b <- featureNames(res.mo) 
b # Names of the spots selected for a  modT-test with an FDR of 0.1
c <- notes(Sim.data)$SpotSig 
c #  Names of the really differentially generated spots

# Compare between t-test and moderate t-test (requires made4 package)
# Just for illustration purpose (only 1 simulation !)
## Not run: 
# cl.t <- comparelists(c,a)
# cl.m <- comparelists(c,b)
# 
# fp <- matrix(c(c(length(a)- (cl.t$Length.Y-length(cl.t$intersect))),
# c(cl.t$Length.Y-length(cl.t$intersect)), 
# c(length(b)- (cl.m$Length.Y-length(cl.m$intersect))),
# c(cl.m$Length.Y-length(cl.m$intersect))),
# nrow=2,ncol=2,byrow=F)
# 
# barplot(fp, main="spots declared significant",names.arg=c("t-test","modT"))
# legend("topleft",c("False Positives","True Positives"), fill=c("grey","black"), bty="n")
# ## End(Not run)