NormRFit-class {normr} | R Documentation |
This S4 class wraps a normR
fit containing counts, fit
configuration and results of the fit. Herein, functions for printing,
summarization and accessing are provided. The
functions enrichR
, diffR
and
regimeR
generate a container of this class to save results of
a normR binomial mixture fitting. Please refer to their documentation for
conventional usage of the normR package.
## S4 method for signature 'NormRFit,character' exportR(x, filename, fdr = 0.01, color = NA, type = c(NA, "bed", "bedGraph", "bigWig")) ## S4 method for signature 'NormRFit,missing' plot(x, y, ...) ## S4 method for signature 'NormRFit' getCounts(x) ## S4 method for signature 'NormRFit' getRanges(x, fdr = NA, k = NULL) ## S4 method for signature 'NormRFit' getPosteriors(x) ## S4 method for signature 'NormRFit' getEnrichment(x, B = NA, F = NA, standardized = TRUE, procs = 1L) ## S4 method for signature 'NormRFit' getPvalues(x, filtered = FALSE) ## S4 method for signature 'NormRFit' getQvalues(x) ## S4 method for signature 'NormRFit' getClasses(x, fdr = NA) ## S4 method for signature 'NormRFit' length(x) ## S4 method for signature 'NormRFit' print(x, digits = max(3L, getOption("digits") - 3L), ...) ## S4 method for signature 'NormRFit' show(object) ## S4 method for signature 'NormRFit' summary(object, print = TRUE, digits = 3, ...)
x |
A |
filename |
A |
fdr |
|
color |
Specified color(s) when printing a bed file. If |
type |
A |
y |
not used. |
... |
optional arguments to be passed directly to the inherited function without alteration and with the original names preserved. |
k |
|
B |
An |
F |
An |
standardized |
A |
procs |
An |
filtered |
A |
digits |
Number of digits to show in number formatting. |
object |
A |
print |
|
When working with instances of this S4 class, it is recommended to only use functions to access contents of this object. Internally, the class holds a map structure of unique elements to reduce memory requirements. #'
getCounts: A list
of length 2 with integer
for control
and treatment each.
getRanges: A GenomicRanges
object.
getPosteriors: A matrix
of posteriors for x@k
mixture
components
getEnrichment: A numeric
of length length(x@n)
giving
the normR computed enrichment.
getPvalues: A numeric
of length length(x@n)
giving
the normR computed Pvalues.
getQvalues: A numeric
of length length(x@filteredT)
giving the FDR-corrected q-values using Storey's method.
getClasses: A integer
specifying assignments of regions to
the mixture model. If x@type == "enrichR"
, it contains 1
for
enriched regions and NA
for non-enriched regions. If x@type ==
"diffR"
, it contains 1
for control-enriched regions, 2
for
treatment-enriched regions and NA
for non-enriched regions. If
x@type == "regimeR"
, it contains >= 1
for regime-enriched
regions and NA
for non-enriched regions.
exportR
: Export results of a normR fit to common file formats.
plot
: Plot a NormRFit.
getCounts
: Get count data for control and treatment.
getRanges
: Get the genomic coordinates of regions analyzed with
information about component assignment.
getPosteriors
: Get computed posteriors for each mixture component.
getEnrichment
: Get normalized enrichment.
getPvalues
: Get normR-computed P-values.
getQvalues
: Get FDR-corrected q-values.
getClasses
: Get component assignments for each region analyzed.
length
: Returns the number of regions analyzed.
print
: Prints a small summary on a NormRFit.
show
: Shows a small summary on a NormRFit.
summary
: Prints a concise summary of a NormRFit.
type
A character
representing the type of fit. One of
c("enrichR","diffR", "regimeR")
.
n
An integer
specifying the number of regions.
ranges
A GenomicRanges
specifying the genomic coordinates of
the regions.
k
An integer
giving the number of binomial mixture components.
B
An integer
specifying the index of the background component.
map
A vector of integer
holding a map to map back
counts
, lnposteriors
, lnenrichment
, lnpvals
,
lnqvals
and classes
. See low level function
normr:::map2uniquePairs
for how the map is generated.
counts
A list
of length two containing a vector of
integer
holding unique counts for control and treatment each. Use
getCounts
to retrieve original count matrix.
amount
A vector of integer
specifying the number of occurences
of each unique control / treatment count pair.
names
A character
of length two specifying the names for
control and treatment.
thetastar
A numeric
giving the calculated naive background
estimation, i.e. sum(getCounts(obj)[2,])/sum(getCounts(obj))
theta
A numeric
of length k
giving the normR fitted
parametrization of k
binomial mixture components.
mixtures
A numeric
of length k
giving the normR fitted
mixture proportions of k
binomial mixture components. Should add up
to one.
lnL
A vector of numeric
holding the log-likelihood-trace of
a normR model fit.
eps
A numeric
used as threshold for normR fit EM convergence.
lnposteriors
A matrix
with length(amount)
rows and
k
columns. It contains the ln posterior probabilities for each unique
control / treatment count pair. Use getPosteriors
to get the
posterior matrix for the original data.
lnenrichment
A numeric
of length length(amount)
holding
calculared normalized enrichment for each unique control / treatment count
pair. The enrichment is calculated with respect to the fitted component
B
. Use getEnrichment
to retrieve enrichment for the
original data.
lnpvals
A numeric
of length length(amount)
holding ln
P-values for each unique control / treatment count pair. Given
theta
of B
the signifcane of enrichment is assigned. Use
getPvalues
to retrieve Pvalues for original data.
thresholdT
An integer
giving the threshold used to filter
P-values for FDR correction. The T-Filter threshold is a calculated
population size for which the null hypothesis (theta
of B
) can
be rejected. eps
specifies the significance level.
filteredT
A vector of integer
giving indices of P-values
passing thresholdT
. Only these P-values will be considered for FDR
correction.
lnqvals
A numeric
of length length(filteredT)
holding
ln q-values (FDR correction). P-values are corrected for multiple testing
using Storey's method.
classes
A integer
of length length(amount)
specifying
the class assignments for each unique control / treatment count pair. These
class assignments are based on the normR model fit. For type ==
"enrichR"
, this vector contains either NA
(not enriched) or 1
(enriched). For type == "diffR"
, this vector contains NA
(unchanged), 1
(differential in ChIP-seq 1) and 2
(differential in ChIP-seq 2). For type == "regimeR"
, this vector
contains NA
(not enriched) and an arbitary number of enrichment class
>= 1
.
Johannes Helmuth helmuth@molgen.mpg.de
normr for function creating this container
require(GenomicRanges) #Create a toy instance of type 'enrichR' fit <- new("NormRFit", type="enrichR", n=10L, ranges=GRanges("chr1", IRanges(seq(1,100,10), width=10)), k=2L, B=1L, map=rep(1:5,2), counts=list(1:5, 1:5), amount=rep(2L,5), names=c("chip", "input"), thetastar=.35, theta=c(.15,.55), mixtures=c(.9,.1), lnL=seq(-50,-1,10), eps=.001, lnposteriors=log(matrix(runif(10), ncol=2)), lnenrichment=log(runif(5,0,.2)), lnpvals=log(runif(5)), filteredT=2:5, thresholdT=1L, lnqvals=log(runif(5,0,.2)), classes=sample(1:2,5,TRUE)) #print some statistics on fits fit summary(fit) ## Not run: #write significant regions to bed #exportR(fit, filename = "enrich.bed", fdr = 0.1) #write normalized enrichment to bigWig #exportR(fit, filename = "enrich.bw") ## End(**Not run**) ###AccessorMethods #get original counts getCounts(fit) #get genomic coordinates for significant ranges as a GenomicRanges instance getRanges(fit, fdr = .1) getPosteriors(fit) getEnrichment(fit) getPvalues(fit) getQvalues(fit) getClasses(fit)