simulationWrapper.filter: Wrapper for simulating M datasets

In jhsiao999/ashbun: Methods and evaluation tools for single cell RNA-seq count data

Description

This different version of simulationWrapper first selects single cell samples, performs filtering, selects a subset of genes, and then performs permutation at the gene-level or sample-level. This is in contrast with the previous approach, where filtering is performed after select a subset of genes and permuting labels.

Usage

simulationWrapper.filter(counts, Nsim = 1, Nsamples = 50, Ngenes = NULL,
  pi0 = NULL, sample_method = c("all_genes", "per_gene"),
  samplesFractionExpressed = 0.25, featuresFractionExpressed = 0.25,
  thresholdDetection = 1, beta_args = args.big_normal(betapi = c(1), betamu
  = c(0), betasd = c(1)))

Arguments

counts	gene by sample count matrix
Nsim	number of simulated datasets
Ngenes	number of genes. Defaults to include all genes in the input data.
Nsample	number of samples per biological condition

Value

List of data generated under three different fractions of null genes. null pi0 = 0. List of Nsim simulated datasets. normal_5 pi0 = 0.5. List of Nsim simulated datasets. For each simulated dataset, I store the count table and a logical vector indicating TRUE = null gene, and FALSE = true DE gene. norma_9 pi0 = 0.9. List of Nsim simulated datasets. List of Nsim simulated datasets. For each simulated dataset, I store the count table and a logical vector indicating TRUE = null gene, and FALSE = true DE gene.

Examples

ipsc_eset <- get(load(system.file("testdata", "HumanTungiPSC.rda", package = "ashbun")))
counts <- exprs(ipsc_eset)[sample(nrow(exprs(ipsc_eset)), ), ]

simdata_list <- simulationWrapper(counts, Nsim = 5, Nsample = 80, Ngenes = 500)

library(singleCellRNASeqHumanTungiPSC)
eset <- HumanTungiPSC
counts <- exprs(eset)[,pData(eset)$individual == "NA19101"]

sim_data_null <- simulationWrapper(counts,
                              Ngenes = 100,
                              Nsamples = 20,
                              sample_method = "all_genes",
                              pi0 = 1)

sim_data_nonnull <- simulationWrapper(counts, Nsim = 2,
                              Ngenes = 100,
                              Nsam = 20,
                              sample_method = "all_genes",
                              pi0 = .5,
                              beta_args = args.big_normal(betapi = 1,
                                                          betamu = 0, betasd = .8))

jhsiao999/ashbun documentation built on May 8, 2019, 11:17 p.m.