npSeq.Simu.Data: Simulate sequencing data

In joey711/npSeq: A non-parametric method for significance analysis of sequencing data

Description

Simulate sequencing data with two class, multiclass, quantitative or survival outcomes.

Usage

npSeq.Simu.Data(dat, seed=10)

Arguments

dat

a list with elements (the first three are required): type: "twoclass", "multiclass", "quant", or "survi". option: "1" for Poisson, "2" for negative binomial with dispersion 0.25, "3" for Poisson with outliers, "4" for negative binomial with outliers. NSAM: number of samples. an integer for quant and survi, and a vector of integers for twoclass and multiclass. NGENE: number of genes. default value 20000. psig: percentage of significant genes. default value 0.3. up.perc: in the significant genes, how many percent are up-regulated. Default value: 0.8.

seed

random seed

Details

This function generate all simulated data for the paper. Different outcome type: two class, multiple class, quantitative, or survival. Different distribution: Poisson, negative binomial, with/without outliers.

Value

a list with all elements in the input dat, and

rmean	gene expression levels.
cmean	sequencing depths.
mu	means of Poission/negative binomial distribution.
y	the outcome vector.
fold.change	the log fold change.
n	the count matrix
delta	TRUE/FALSE indicating whether a gene is differentially expressed.
gamma	for survival data. observed (1) or censored (0).

Author(s)

Jun Li

References

Jun Li and Robert Tibshirani (2011). Finding consistent patterns: a nonparametric approach for identifying differential expression in RNA-Seq data. To appear, Statistical Methods in Medical Research.

Jun Li, Daniela M. Witten, Iain Johnstone, Robert Tibshirani (2011). Normalization, testing, and false discovery rate estimation for RNA-sequencing data. To appear, Biostatistics.

Examples

## two class Poisson-distributed data with 12 samples in each class
dat <- npSeq.Simu.Data(list(type='twoclass', option=1, NSAM=c(12, 12)))

## two class negative binomial-distributed data with outliers,
## 12 samples in each class
dat <- npSeq.Simu.Data(list(type='twoclass', option=4, NSAM=c(12, 12)))

## 4 class Poisson-distributed data with outliers,
## 6 samples in each class
dat <- npSeq.Simu.Data(list(type='multiclass', option=3, NSAM=c(6, 6, 6, 6)))

## quantitative negative binomial-distributed data with outliers,
## 24 samples totally
dat <- npSeq.Simu.Data(list(type='quant', option=4, NSAM=24))

## survival negative binomial-distributed data with outliers,
## 24 samples totally
dat <- npSeq.Simu.Data(list(type='survi', option=4, NSAM=24))

joey711/npSeq documentation built on May 19, 2019, 3:01 p.m.