| Title: | OSAT: Optimal Sample Assignment Tool |
|---|---|
| Description: | A sizable genomics study such as microarray often involves the use of multiple batches (groups) of experiment due to practical complication. To minimize batch effects, a careful experiment design should ensure the even distribution of biological groups and confounding factors across batches. OSAT (Optimal Sample Assignment Tool) is developed to facilitate the allocation of collected samples to different batches. With minimum steps, it produces setup that optimizes the even distribution of samples in groups of biological interest into different batches, reducing the confounding or correlation between batches and the biological variables of interest. It can also optimize the even distribution of confounding factors across batches. Our tool can handle challenging instances where incomplete and unbalanced sample collections are involved as well as ideal balanced RCBD. OSAT provides a number of predefined layout for some of the most commonly used genomics platform. Related paper can be find at http://www.biomedcentral.com/1471-2164/13/689 . |
| Authors: | Li Yan |
| Maintainer: | Li Yan <[email protected]> |
| License: | Artistic-2.0 |
| Version: | 1.53.0 |
| Built: | 2024-07-03 05:47:27 UTC |
| Source: | https://github.com/bioc/OSAT |
"BeadChip"
A predefined class represent BeadChip from Illumina. Inherited from
class gSlide.
IlluminaBeadChip: A slide with 6 rows and 2 columns. The 12
wells is numbered columns first.
GenotypingChip: A slide with 12 rows and one column.
Class "gSlide", directly.
Class "gArray", by class "gSlide", distance 2.
signature(x = "BeadChip"): Return a data
frame represent the layout of the chip.
newChip <- new("BeadChip", nRows=6, nColumns=1, byrow=FALSE, comment="mock chip") newChip GenotypingChip IlluminaBeadChipnewChip <- new("BeadChip", nRows=6, nColumns=1, byrow=FALSE, comment="mock chip") newChip GenotypingChip IlluminaBeadChip
"BeadPlate"
A class represent a plate consistent of multiple chips. Inherite from
class gPlate.
Objects can be created by calls of the form new("BeadPlate", chip, nRows, nColumns, comment, ...).
chip:Object of class "gSlide". The slide used
in the plate.
nRows:Object of class "integer". Number of
clide per row.
nColumns:Object of class "integer". Number of
slide per column.
layout:Object of class "data.frame". A data frame represent
the layout of the plate.
byrow:Object of class "logical". Order of
slide on the plate.
metadata:Object of class "list".A brief description of the plate.
IlluminaBeadChip96Plate
IlluminaBeadChip48Plate
IlluminaBeadChip24Plate
are plates that hold 2, 4, 8 IlluminaBeadChip chips and have 24, 48, 96 wells, respectively.
Class "gPlate", directly.
Class "gArray", by class "gPlate", distance 2.
library("OSAT") newPlate <- new("BeadPlate", chip=IlluminaBeadChip, nRows=2L, nColumns=4L, comment="mock plate") newPlatelibrary("OSAT") newPlate <- new("BeadPlate", chip=IlluminaBeadChip, nRows=2L, nColumns=4L, comment="mock plate") newPlate
Create a optimized sample assignment.
create.optimized.setup(fun = "default", sample, container, ...)create.optimized.setup(fun = "default", sample, container, ...)
fun |
The name of the optimization function. When omitted, it is the same
as |
sample |
A |
container |
A |
... |
Additional parameters passed to optimization function. |
Currently two methods are
available: optimal.shuffle (default) and optimal.block
methods. The function is equivalent to create.experiment.setup()
followed by corresponding optimization function of the same name.
A gExperimentSetup object is returned to store all related information.
library("OSAT") # data as an example inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") ) gs gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') gc # demonstration only. nSim=5000 or more are commonly used. gSetup <- create.optimized.setup(sample=gs, container=gc, nSim=500)library("OSAT") # data as an example inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") ) gs gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') gc # demonstration only. nSim=5000 or more are commonly used. gSetup <- create.optimized.setup(sample=gs, container=gc, nSim=500)
Data objects created using the example data for demonstration.
data(example.setup)data(example.setup)
See vignette for codes that created these data objects.
a data frame created from example sample information text file.
a class gSample object created based on example
sample using function setup.sample()..
a class gContainer object using function setup.container().
a class gExperimentSetup object. Created using
optimal.shuffle() method.
a class gExperimentSetup object. Created using
optimal.block() method.
library("OSAT") data(example.setup) head(pheno) gs gc gSetuplibrary("OSAT") data(example.setup) head(pheno) gs gc gSetup
"gArray"
The virtual class gArray is used to stadarize the storage and access of array-like experiment arrangement.It represent the arrangement of the experiment units, such as slide ans plate.
A virtual Class: No objects may be created from it.
nRows:Object of class "integer". Number of
units per row.
nColumns:Object of class "integer". Number of
units per column.
layout:Object of class "data.frame". A data
frame represent the layout and order of postiion.
byrow:Object of class "logical". The order of
position is filled by row or not. Default is FALSE, same as
used in mactrix().
metadata:Object of class "list". A list that
holds brief description of the class and other information.
In the following code snippets, x is an gArray object.
signature(x = "gArray"): Return the dimmension of
the array.
signature(x = "gArray"): Return a data frame
represent the layout postions of the array.
signature(x = "gArray"): Set the metadata.
signature(x = "gArray"): Get the metadata.
signature(x = "gArray"): Return the number of columns.
signature(x = "gArray"): Return the number of rows.
"gAssembly"
A class represent an assembly of experiment plates.
Objects can be created by calls of the form new("gAssembly", plate, n, comment).
plate:Object of class "gArray". Plate used in
the experiment.
n:Object of class "integer". NUmber of plates used.
layout:Object of class "data.frame". The
layout of wells in the assembly.
metadata:Object of class "list". A list holds
meta data of the assembly.
signature(x = "gAssembly"): Return a data
frame that represent the wells location in the assembly.
signature(x = "gAssembly"): Get metadata.
signature(x = "gAssembly"): Set metadata.
signature(object = "gAssembly"): Print a brief
summary of the assembly.
"gContainer".
"gContainer"
A class for storage information related to the experiment container setup.
setup.container(plate, n, batch = "plates", exclude = NULL)setup.container(plate, n, batch = "plates", exclude = NULL)
plate |
A object of |
n |
Number of plates used in the experiment. |
batch |
The level where batch effect is considered. |
exclude |
A data frame indicate location of wells in the container that should be excluded from sample assigment in the experiment. |
A gContainer object.
Objects can be created by calls of the form
setup.container(plate, n, batch, exclude).
plate:Object of class "gPlate". Plate used in
the experiment.
n:Object of class "integer". Number of plate used.
batch:Object of class "character". On what
level batch effect are considered. Could be "plates" or
"chips".
exclude:Object of class "data.frame". A data
frame indicate wells that should be excluded from the sample Assignment.
data:Object of class "list". A list holds
summaries and other useful informaiton.
metadata:Object of class "list". A list for a brief
description and other useful informtion.
signature(x = "gContainer"): Return the
assembly of the plates used in the experiments.
signature(x = "gContainer"): Return a data
frame that holds the layout of available wells in the container.
signature(object = "gContainer"): Exclude some
wells/chips/plates from the container for randomization.
signature(x = "gContainer"): Set the metadata.
signature(x = "gContainer"): Get the metadata.
signature(object = "gContainer"): A brief summary
of the container.
library("OSAT") # a container consist of 6 predefined Illumina plates with 96 wells each gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') gc # to exclude first wells on first chips of each plate. (excludedWells <- data.frame(plates=1:6, chips=rep(1,6), wells=rep(1,6)) ) gc3 <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates', exclude=excludedWells)library("OSAT") # a container consist of 6 predefined Illumina plates with 96 wells each gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') gc # to exclude first wells on first chips of each plate. (excludedWells <- data.frame(plates=1:6, chips=rep(1,6), wells=rep(1,6)) ) gc3 <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates', exclude=excludedWells)
Return a data frame that contains sample and assigned well postion.
get.experiment.setup(x)get.experiment.setup(x)
x |
A |
A data frame is returned in the order of initial sample data frame, with additional columns indicate assgined well position.
library("OSAT") # data as an example inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") ) gs gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') gc # demonstration only. nSim=5000 or more are commonly used. gSetup <- create.optimized.setup(sample=gs, container=gc, nSim=500) mySetup <- get.experiment.setup(gSetup)library("OSAT") # data as an example inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") ) gs gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') gc # demonstration only. nSim=5000 or more are commonly used. gSetup <- create.optimized.setup(sample=gs, container=gc, nSim=500) mySetup <- get.experiment.setup(gSetup)
"gExperimentSetup"
A class stores all relevant informations related to the experimantal sample assgnment.
Function create.experiment.setup() create a block randomized
experiment assignment, without optimization step.
Functions optimal.shuffle() and optimal.block optimize
setup using different optimization methods.
Functiion create.optimized.setup() create a optimized setup directly.
expSetup:Object of class "data.frame". A data
frame represents the sample placement to well locations in the container.
data:Object of class "list". A list that hold
a gSample object, a gContainer object and assignment
link the two objects.
summaryInfo:Object of class "list". Some
summary of the sample, container and assignment.
metadata:Object of class "list". Metadata of
the object.
signature(x = "gExperimentSetup"): Get the chip/plate assembly used for container.
signature(x = "gExperimentSetup"): Return the
gSample object.
: ...
signature(x = "gExperimentSetup"): Get the metadata.
signature(x = "gExperimentSetup"): Set the metadata.
signature(x = "gExperimentSetup", y =
"MSAroboticPlate"): A method map the experiment setup to a set of MSA 96
wells robotic plates.
signature(x = "gExperimentSetup", y = "missing"):
Visual
signature(object = "gExperimentSetup"): ...
signature(object = "gExperimentSetup"): ...
: Return a data frame with linked sample and container information.
library("OSAT") # data as an example inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") ) gs gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') gc gSetup0 <- create.experiment.setup(sample=gs, container=gc)library("OSAT") # data as an example inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") ) gs gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') gc gSetup0 <- create.experiment.setup(sample=gs, container=gc)
"gPlate"
A virtual class represent a experiment plate that is consistent of multiple slides.
A virtual Class: No objects may be created from it.
chip:Object of class "gSlide". A virtual class represent a experiment plate that is consistent of multiple slides.
nRows:Object of class "integer". Number of
slide per row.
nColumns:Object of class "integer". Number of
slide per column.
layout:Object of class "data.frame". A data
frame represent the layout of the plate. Usually down to the wells
level.
byrow:Object of class "logical". The order of
the position is filled by row or not. Default is FALSE, same as
used in mactrix().
metadata:Object of class "list". A list that
holds brief description of the class and other information.
Class "gArray", directly.
gSample
Create a class used for storage of sample related information.
setup.sample(x, optimal, strata) summary(object, ...)setup.sample(x, optimal, strata) summary(object, ...)
x |
a data frame holds sample variables. |
optimal |
a vector of sample varible names to be treated as optimal variables. |
strata |
a vector of sample varible names to be treated as block
variables. If omitted, the first element of |
object |
an object of class |
... |
additional arguments affecting the summary produced. |
An object of gSample class for function setup.sample().
A list of two tables for function summary(object):
strataTable |
Sample freqency table by stata variables. |
optimalTable |
Sample freqency table by optimal variables. |
rawData:Object of class "data.frame" The
original data frame that holds the sample informaiton.
optimal:Object of class "character" A
charactor vector of column names in the sample info data
frame. Represent the optimal variables.
strata:Object of class "character" A
charactor vector of column names in the sample info data
frame. Represent the blocking variables.
data:Object of class "list" A list holds
summaries and other useful information.
signature(object = "gSample"): ...
inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") )inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") )
"gSlide"
A class repressent a slide used in experiment. Inherited from
Class "gArray". The layout of the slide is descriped by a data
frame with columns of "rows", "columns", and "wells".
Objects can be created by calls of the form new("gSlide", nRows, nColumns, byrow, comment).
Class "gArray", directly.
signature(x = "gSlide"): ...
signature(.Object = "gSlide"): ...
signature(object = "gSlide"): ...
newSlide <- new("gSlide", nRows=2, nColumns=2, byrow=FALSE,comment="mock slide") newSlidenewSlide <- new("gSlide", nRows=2, nColumns=2, byrow=FALSE,comment="mock slide") newSlide
"MSAroboticPlate"
A class store layout information of MSA robotic loader plate.
Class "gArray", directly.
signature(x = "data.frame", y =
"MSAroboticPlate"): Return a data frame that assign samples
(represented by the input data frame) to MSA robotic plate sequentially.
signature(x = "gExperimentSetup", y =
"MSAroboticPlate"): Return a data frame that assign samples from an
gExperimentSetup object to MSA robotic plate.
signature(object = "MSAroboticPlate"): Shows the
layout of the plate.
MSA4.plate: A predeined object of class MSAroboticPlate
that represent a 96 position plate.
BeadChip96ToMSA4MAPMap: The loading order a MSA4 robotic loader
used to load BeadChips.
library("OSAT") # data as an example inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") ) gs gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') gc gSetup <- create.optimized.setup(sample=gs, container=gc, nSim=100) out <- map.to.MSA(gSetup, MSA4.plate)library("OSAT") # data as an example inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") ) gs gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') gc gSetup <- create.optimized.setup(sample=gs, container=gc, nSim=100) out <- map.to.MSA(gSetup, MSA4.plate)
Plot multiple bar plots based on a single data frame.
multi.barplot(x, grpVar = "plates", varList, main = NULL, ...)multi.barplot(x, grpVar = "plates", varList, main = NULL, ...)
x |
A data frame. |
grpVar |
The variable for x-axix. |
varList |
A vector of variables, each will be used as y-axis for the multiple bar plot. |
main |
The overall title of the plot. |
... |
Additinal parameters for the plot function. |
Use the side effect for plot. No return value.
## create a random assignment and check it library("OSAT") data(example.setup) set.seed(10) c1 <- getLayout(gc) # available wells c1 <- c1[order(runif(nrow(c1))),] # shuffle randomly ranomSetup <- cbind(pheno, c1[1:nrow(pheno), ]) # create a sample assignment multi.barplot(ranomSetup, grpVar='plates', varList=c("SampleType", "Race", "AgeGrp"), main="A random case")## create a random assignment and check it library("OSAT") data(example.setup) set.seed(10) c1 <- getLayout(gc) # available wells c1 <- c1[order(runif(nrow(c1))),] # shuffle randomly ranomSetup <- cbind(pheno, c1[1:nrow(pheno), ]) # create a sample assignment multi.barplot(ranomSetup, grpVar='plates', varList=c("SampleType", "Race", "AgeGrp"), main="A random case")
Do a few Chi-squre tests based on the same data frame.
multi.chisq.test(x, grpVar = "plates", varList, main = NULL)multi.chisq.test(x, grpVar = "plates", varList, main = NULL)
x |
A data frame. |
grpVar |
Common variables. Default is 'plate'. |
varList |
A vector of variables. |
main |
The overall title. |
## create a random assignment and check it library("OSAT") data(example.setup) set.seed(10) c1 <- getLayout(gc) # available wells c1 <- c1[order(runif(nrow(c1))),] # shuffle randomly ranomSetup <- cbind(pheno, c1[1:nrow(pheno), ]) # create a sample assignment multi.chisq.test(ranomSetup, grpVar='plates', varList=c("SampleType", "Race", "AgeGrp"))## create a random assignment and check it library("OSAT") data(example.setup) set.seed(10) c1 <- getLayout(gc) # available wells c1 <- c1[order(runif(nrow(c1))),] # shuffle randomly ranomSetup <- cbind(pheno, c1[1:nrow(pheno), ]) # create a sample assignment multi.chisq.test(ranomSetup, grpVar='plates', varList=c("SampleType", "Race", "AgeGrp"))
Optimize a sample assingment setup by selecting from multiple candidate setup.
optimal.block(x, nSim = 100)optimal.block(x, nSim = 100)
x |
A |
nSim |
Number of candidate setup created. |
Multiple (typically thousands of or more) sample assignment setups are first generated, based only on the list of specified blocking variable(s). Then, the optimal setup is chosen by selecting the setup of sample assignment (from the pool generated in blocking step) which minimizes the value of the objective function based on all variables considered.
A gExperimentSetup object, after optimization.
library("OSAT") # data as an example inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") ) gs gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') gc gSetup0 <- create.experiment.setup(sample=gs, container=gc) g2 <- optimal.block(gSetup0, nSim=100)library("OSAT") # data as an example inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") ) gs gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') gc gSetup0 <- create.experiment.setup(sample=gs, container=gc) g2 <- optimal.block(gSetup0, nSim=100)
Optimize a sample assingment setup by shuffling.
optimal.shuffle(x, nSim, k)optimal.shuffle(x, nSim, k)
x |
A |
nSim |
Number of shuffling steps. |
k |
Number of samples been shuffled. Default |
Given any gExperimentSetup object, we randomly select k samples from different batches and shuffle them between batches to create a new sample assignment. k = 2 by default but could be any number up to half of the sample size. Value of the objective function is calculated on the new setup and compared to that of the original one. If the value is smaller then the new assignment replaces the previous one. This procedure will continue until we reach a preset number of attempts (usually in the tens of thousands).
A class gExperimentSetup object, after optimized.
library("OSAT") # data as an example inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") ) gs gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') gc gSetup0 <- create.experiment.setup(sample=gs, container=gc) # demonstration only. nSim=5000 or more are commonly used. g3 <- optimal.shuffle(gSetup0, nSim=500, k=2)library("OSAT") # data as an example inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") ) gs gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') gc gSetup0 <- create.experiment.setup(sample=gs, container=gc) # demonstration only. nSim=5000 or more are commonly used. g3 <- optimal.shuffle(gSetup0, nSim=500, k=2)
plot
Create bar plots based on the experiment setup.
signature(x = "gExperimentSetup", y = "missing")Plot distribution of relevant variables over different batches.
Show predefined objects in the package.
predefined()predefined()
Currently layout of two chips, three IlluminaBeadChip plates, and the MSA4 robotic loader are defined.
Provide a visual summary of the sample placement per plates, and Chi-squre test of dependence between plates and other considered variables from sample.
QC(object, main = NULL, ...)QC(object, main = NULL, ...)
object |
An object of class |
main |
Mail title on the bar plot. |
... |
Additional plot parameters. |
library("OSAT") inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") ) ## create object that represents the used in the experiment gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') ## create an optimized setup. # demonstration only. nSim=5000 or more are commonly used. gSetup <- create.optimized.setup(sample=gs, container=gc, nSim=500) QC(gSetup)library("OSAT") inPath <- system.file("extdata", package="OSAT") pheno <- read.table(file.path(inPath, 'samples.txt'), header=TRUE, sep="\t") ## create object to hold sample information gs <- setup.sample(pheno, optimal=c("SampleType", "Race", "AgeGrp"), strata=c("SampleType") ) ## create object that represents the used in the experiment gc <- setup.container(IlluminaBeadChip96Plate, 6, batch='plates') ## create an optimized setup. # demonstration only. nSim=5000 or more are commonly used. gSetup <- create.optimized.setup(sample=gs, container=gc, nSim=500) QC(gSetup)