EDDA (version 1.10.0)

generateData: generate count data

Description

Simulate count data using different models and settings.

Usage

generateData(SimulModel="Full", SampleVar="medium", 
		ControlRep=5, CaseRep=ControlRep, EntityCount=1000, FC="Norm(2,1)",
		perDiffAbund=0.1, upPDA=perDiffAbund/2, downPDA=perDiffAbund/2,
		numDataPoints=100, AbundProfile = "HBR", modelFile = NULL, minAbund=10,varLibsizes=0.1, outlier=FALSE,perOutlier=0.15, factorOutlier=100,
		inputCount=NULL,inputLabel=NULL,SimulType="auto")

Arguments

SimulModel

Simulation model used. Default is "Full". SimulModel="NegBinomial" is negtive binomial model which generates data using negtive binomial distribution. SimulModel="Multinom" is multinomial model which generates data mimicking the multinomial sampling process. SimulModel="Full" is a model which combines "NegBinomial" and "Multinom". SimulModel="ModelFree" uses model free approach to generate data by sub-sampling counts from modelFile (if modelFile != NULL) or from input File (if modelFile == NULL)

SampleVar

Sample variation: Default is "medium". It could be "low", "medium" and "high" or a real number.

ControlRep

Number of replicates for control group. Default is 5.

CaseRep

Number of replicates for case group. Default is same as ControlRep.

EntityCount

Entity count. Default is 1000.

FC

Fold change type. It can be "Norm(mu,sigma)", "logNorm(mu,sigma)", "log2Norm(mu,sigma)" or "Unif(a,b)". mu,sigma and a,b need be predefined. Default is "Norm(2,1)".

perDiffAbund

Percentage of differential abundance. Default is 0.1 (i.e "10 percent").

upPDA

Percentage of up-regulated differential abundance. Default is perDiffAbund/2.

downPDA

Percentage of down-regulated differential abundance. Default is perDiffAbund/2.

numDataPoints

Number of data points. Default is 100.

AbundProfile

AbundProfile for average abundance profile. It can be either the different profiles used in the paper ("HBR", "BP" and "Wu") or it can be location of the abundance profile. Default is "HBR".

modelFile

Sample data file for model free approach. Default is NULL. If modelFile = NULL, Model Free approach will subsample from Input file. If modelFile = "SingleCell", Model Free approach will subsample from the available single cell RNA-seq data. if modelFile is the name of a count file, this count file will be used as sample file for sub-sampling.

minAbund

Minimum abundance cutoff. Default is 10.

varLibsizes

Variability between library sizes. Default is 0.1.

outlier

Outlier model. Default is FALSE, i.e. outlier model is turned off.

perOutlier

Percentage of added outliers. Default is 0.15.

factorOutlier

Scaling factor to generate outliers. Default is 100.

inputCount

Input count file. Default is NULL. If not NULL, it learns the parameters (modelFile, SampleVar, perDiffAbund, upPDA, and downPDA) from count data.

inputLabel

Label of input count file. The label should be sequence of 0 or 1. Default is NULL.

SimulType

Simulation type. It is used only when user has pilot data. Default is "auto". SimulType = "auto", all the parameters (EntityCount, ControlRep, CaseRep, numDataPoint and others) are learned from user's pilot data. SimulType = "auto1", ControlRep, CaseRep, numDataPoint are specified by user input; while EntityCount and all others are learned from user's pilot data. SimulType = "auto2", EntityCount, ControlRep, CaseRep, numDataPoint are specified by user input; while all others are learned from user's pilot data.

Value

count

Count matrix.

DiffAbundList

Differential abundance list.

dataLabel

Data label.

References

Luo Huaien, Li Juntao,Chia Kuan Hui Burton, Shyam Prabhakar, Paul Robson, Niranjan Nagarajan, The importance of study design for detecting differentially abundant features in high-throughput experiments, under review.

Examples

# generate data with all default options.
data <- generateData()
dim(data$count)
dim(data$DiffAbundList)
data$dataLabel

# generate data with input count.
x <- matrix(rnbinom(1000*15,size=1,mu=10), nrow=1000, ncol=15);
x.lable=c(rep(0,10),rep(1,5))
x[1:50,11:15] <- x[1:50,11:15]*10
x.name=paste("g",1:1000,sep="");
write.table(cbind(x.name,x),"count.txt",row.names =FALSE, sep ='\t')

data <- generateData(inputCount="count.txt",inputLabel=x.lable)
dim(data$count)
dim(data$DiffAbundList)
data$dataLabel

# or generate data with input count and redefined parameters.
data <- generateData(inputCount="count.txt",inputLabel=x.lable,
                     ControlRep=10,CaseRep=10,EntityCount=3000,SimulType="auto2")
dim(data$count)
dim(data$DiffAbundList)
data$dataLabel