gendesign: Statistical design for RNA-seq experiments aimed at...

In Lina-Gao/DEdesign: Statistical design for RNA-seq experiments aimed at identifying differentially expressed genes

Description

Generates block design to assign samples to sequencing lanes and adapters for Illumina flow cell. The gendesign function is the main function of the DEdesign package. It calls the design function from blocksdesign package (by Dr. Rodney Edmondson), which can construct nested and crossed block designs for factorial treatments.

Usage

gendesign(treatments, nperlane = 4, search.surrounding = 0, seed = 1,
  searches = NULL)

Arguments

treatments	a data.frame showing number of samples per treatment group. For single factor experiment, treatments should contain two columns with one column showing levels of the factor and the other column named "replicates" for number of samples in each level. For factorial experiments, there should be one column for each factor, and the "replicates" column gives number of samples in each combination of factor levels. Please see examples for a template. For the purpose of plotting, please use short factor names and in factorial design using numbers for factor levels is recommended.
nperlane	An integer for number of samples per lane for sequencing. Default is 4. This is used to generate Design in results.
search.surrounding	An non-negative integer with default 0, which means to only find design for number of samples per lane defined by nperlane. For other values, designs using min(3,(nperlane-search.surrounding)) to (nperlane+search.surrounding) samples per lane will be compared to give suggestedDesign. suggestedDesign is the design with minimal number of flowcells among all candidate designs that gives the highest lane block efficiency (when more than one designs meet these criteria, design with highest nperlane is selected).
seed	an integer initializing the random number generator. The default is seed=1. Designs can be rebuilt repeatedly using different seed to check that a near-optimum design has been found.
searches	the maximum number of local optima searched at each stage of a treatment and block design optimization. The default depends on the design size. For optimum results, try large number of searches.

Value

gendesign returns an object of class "DEdesign", which is a list containing the following components:

input A list showing input parameters to the function

Design A list with two elements: design and BlocksEfficiency. design is a data frame giving flowcell, lane and adapter assignment for treatment groups. BlocksEfficiency is a data frame giving block efficiencies (D-Efficiencies) for lane and adapter (and/or flowcell when applicable).

suggestedDesign When search.surrounding >0, a list with two elements are returned here: design and BlocksEfficiency. design is a data frame giving flowcell, lane and adapter assignment for treatment groups. BlocksEfficiency is a data frame giving block efficiencies (D-Efficiencies) for lane and adapter (and/or flowcell when applicable). Designs using min(3,(nperlane-search.surrounding)) to (nperlane+search.surrounding) samples per lane will be compared to give suggestedDesign. suggestedDesign is the design with minimal number of flowcells among all candidate designs that gives the highest lane block efficiency (when more than one designs meet these criteria, design with highest nperlane is selected).

Examples

gendesign(treatments= data.frame(trt = letters[1:4], replicates = rep(4,4)))

treatments <- data.frame(expand.grid(A=factor(1:2), B=factor(1:5)),replicates = 2)
des <- gendesign(treatments=treatments, nperlane=4, search.surrounding = 2)
designDF(des,selection="Design")
designDF(des,selection="suggestedDesign")
efficiency(des,selection="Design")
efficiency(des,selection="suggestedDesign")

Lina-Gao/DEdesign documentation built on May 5, 2019, 2:39 a.m.