madsim {madsim}

Main function of the package "madsim"

Description

function madsim() allows to generate two biological conditions synthetic microarray dataset with known characteristics. These data have similar behavior as those obtained with current microarray platforms. Hence, they can be used for performance evaluation of data meta-analysis methods.

Usage

madsim(mdata = NULL, n = 10000, ratio = 0,
              fparams = data.frame(m1=7,m2=7,shape2=4,lb=4,ub=14,pde=0.02,sym=0.5),
              dparams = data.frame(lambda1=0.13, lambda2=2, muminde=1, sdde=0.5),
              sdn = 0.4, rseed = 50)

Arguments

mdata

a data frame with numerical values to be used as seed, its length should be greater than 100. When set to NULL (default) data generated are fully synthetic: mdata = NULL

n

an integer specifying the number of genes in the data generated: n = 10000

ratio

a flag (0,1) allowing to have log2 intensitie or log2 ratio: ratio = 0

fparams

a data frame containing 7 components defining the data lower (lb) and upper bound (ub), the beta distribution shape (shape2) parameter, the percentage of differentially expressed (pde) number of genes and the partition of the number of down and up regulated (sym) genes:
fparams=data.frame(m1=7,m2=7,shape2=2,lb=4,ub=14,pde=0.02,sym=0.5)

dparams

a data frame containing 4 components defining how low and high expressed genes are distributed (lambda1), and how changes are for DE genes (lambda2, muminde, sdde):
dparams = data.frame(lambda1=0.13,lambda2=2,muminde=1,sdde=0.5)

sdn

a positive scalar used as standard deviation for the additive gaussian noise: sdn = 0.4

rseed

an integer used as seed for generating random number by the computer in use: rseed = 50

Details

User provides a subset of parameters. A detailed description of these parameters is available in the reference given below. Default parameters settings (in arguments above) can be modified.

Value

Returned is a data frame containing 3 components

xdata

a dataset with sizes, the number of rows and columns, specified by input parameters n and m1+m2, respectively

xid

a vector of indexes with values are from the set (0, -1, 1). These values are used for non differentially expressed, down- and up-regulated genes

xsd

a scalar containing the standard deviation of first column of the dataset generated

Author(s)

Doulaye Dembele

References

Dembele D. (2013), A Flexible Microarray Data Simulation Model. Microarrays, 2013, 2(2):115-130

Examples

    # load a sample of real microarray data
    data(madsim_test)

    # set parameters settings
    mdata <- madsim_test$V1;
    fparams <- data.frame(m1 = 7, m2 = 7, shape2 = 4, lb = 4, ub = 14,pde=0.02,sym=0.5);
    dparams <- data.frame(lambda1 = 0.13, lambda2 = 2, muminde = 1, sdde = 0.5);
    sdn <- 0.4; rseed <- 50;

    # generate fully synthetic data
    mydata1 <- madsim(mdata = NULL, n = 10000, ratio = 0, fparams, dparams, sdn, rseed);

    # use true affymetrix data to generate synthetic data
    mydata2 <- madsim(mdata = madsim_test, n=10000, ratio=0,fparams,dparams,sdn,rseed);

    
    A1 <- 0.5*(mydata1$xdata[,12] + mydata1$xdata[,1]);
    M1 <- mydata1$xdata[,12] - mydata1$xdata[,1];

    A2 <- 0.5*(mydata2$xdata[,12] + mydata2$xdata[,1]);
    M2 <- mydata2$xdata[,12] - mydata2$xdata[,1];

    # draw MA plot using samples 1 and 12
    op <- par(mfrow = c())
       plot(A1,M1)
       plot(A2,M2)
    par(op)