R/ScpModel-Class.R
ScpModel-class.RdAn ScpModel object must be always stored in the metadata() of
an object that inherits from the SingleCellExperiment class. The
ScpModel object should never be accessed directly by the
user. Instead, we provide several setter function to retrieve
information that may be useful to the user.The ScpModel class
contains several slots:
scpModelFormula: a formula object controlling which
variables are to be modelled.
scpModelInputIndex: a numeric(1), selecting the assay to use
in the SingleCellExperiment object as input matrix. Note that
this slot serves as a pointer, meaning that the quantitative
data is not duplicated. Any change to the assay in the
SingleCellExperiment will impact the estimation of the
ScpModel object.
scpModelFilterThreshold: A numeric(1) indicating the minimal
n/p ratio required for a feature to be included in further model
exploration. n/p is the number of measured values for a features
divided by the number of coefficients to estimate. n/p cannot be
smaller than 1 because this would lead to over-specified models.
scpModelFitList: A List that contains the model results for
each feature. Each element is a ScpModelFit object (see
ScpModelFit)
scpModelFormula(object, name)
scpModelInput(object, name, filtered = TRUE)
scpModelFilterThreshold(object, name)
scpModelFilterNPRatio(object, name, filtered = TRUE)
scpModelResiduals(object, name, join = TRUE, filtered = TRUE)
scpModelEffects(object, name, join = TRUE, filtered = TRUE)
scpModelNames(object)
scpModelFilterThreshold(object, name) <- valueAn object that inherits from the
SingleCellExperiment class.
A character(1) providing the name to use to store or
retrieve the modelling results. When retrieving a model and
name is missing, the name of the first model found in
object is used.
A logical(1) indicating whether the output
should return all features (FALSE) or the features that
comply to the n/p ratio threshold (TRUE).
A logical(1) indicating whether the output should be
combined in a single matrix (TRUE) or it should be returned
as a list with one element for each feature (FALSE). When
TRUE, any gaps across features will be filled with NA's.
An numeric(1), the new value for the n/p ratio
threshold
Each slot has a getter function associated:
scpModelNames(): returns a vector of names of ScpModel
objects stored in the SingleCellExperiment object.
scpModelFormula(): returns the formula slot of the ScpModel
within an object that inherits from the SummarizedExperiment
class.
scpModelFilterThreshold(): returns the n/p ration threshold
used for feature filtering.
scpModelInput(): returns a matrix with the quantitative
values used as input of the model. Hence, the matrix contains
the data before modelling. If filtered = TRUE, the feature of
the matrix are restricted to the features that satisfy the n/p
ratio threshold.
scpModelFilterNPRatio(): returns the computed n/p ratio for
each feature. If filtered = TRUE, the function returns only
the n/p of the features that satisfy the n/p ratio threshold.
scpModelResiduals(): when join = FALSE, the function returns
a list where each element corresponds to a feature and contains
the estimated residuals. When join = TRUE (default), the function
combines the list into a matrix with features in rows and cells
in columns, and filling the gaps with NA. If filtered = TRUE,
the feature of the matrix are restricted to the features that
satisfy the n/p ratio threshold.
scpModelEffects(): when join = FALSE, the function return a
list where each element of the list corresponds to a feature.
Each element contains another list with as many elements as
variable in the model and each element contains the data effect vector
for that vector. When join = TRUE (default), each element of the list is
a matrix with features in rows and cells in columns where gaps
are filled with NA. If filtered = TRUE, the feature of the
matrix are restricted to the features that satisfy the n/p
ratio threshold.
Setter:
scpModelFilterThreshold<-(): the function changes the n/p
ratio threshold used for filtering features.
ScpModelFit for a description of the class that store modelling results
ScpModel-Workflow that uses the class to store the estimated model.
data("leduc_minimal")
####---- Getters ----####
scpModelNames(leduc_minimal)
#> [1] "model"
scpModelFormula(leduc_minimal)
#> ~1 + Channel + Set + MedianIntensity + SampleType
dim(leduc_minimal)
#> [1] 200 73
dim(scpModelInput(leduc_minimal))
#> [1] 140 73
dim(scpModelInput(leduc_minimal, filtered = FALSE))
#> [1] 200 73
head(scpModelFilterNPRatio(leduc_minimal))
#> SAVEDEGLK APNVVVTR IVVVTAGVR GFQEVVTPNIFNSR ENAYDLEANLAVLK
#> 1.277778 3.476190 1.411765 2.333333 1.235294
#> IGPLGLSPK
#> 1.888889
dim(scpModelResiduals(leduc_minimal))
#> [1] 140 73
dim(scpModelResiduals(leduc_minimal, filtered = FALSE))
#> [1] 200 73
scpModelResiduals(leduc_minimal, join = FALSE)
#> NumericList of length 140
#> [["SAVEDEGLK"]] wAL00284RI5=0.699508268833165 ...
#> [["APNVVVTR"]] eAL00219RI5=-0.211268540934462 ...
#> [["IVVVTAGVR"]] wAL00284RI5=0.103674283809887 ...
#> [["GFQEVVTPNIFNSR"]] eAL00219RI5=0.637577960221826 ...
#> [["ENAYDLEANLAVLK"]] eAL00220RI5=0.340761402427571 ...
#> [["IGPLGLSPK"]] eAL00220RI5=-0.111917031828062 ...
#> [["VAETANEEEVKK"]] eAL00219RI5=-0.0862785009338332 ...
#> [["TADDPSLSLIK"]] eAL00220RI5=0.0148848162385562 ...
#> [["GPPPSYGGSSR"]] eAL00219RI5=-0.117374478914073 ...
#> [["IDATSASVLASR"]] eAL00219RI5=-0.145390183837019 ...
#> ...
#> <130 more elements>
scpModelEffects(leduc_minimal)
#> List of length 4
#> names(4): Channel Set MedianIntensity SampleType
dim(scpModelEffects(leduc_minimal)$Set)
#> [1] 140 73
dim(scpModelEffects(leduc_minimal, filtered = FALSE)$Set)
#> [1] 200 73
scpModelEffects(leduc_minimal, join = FALSE)[[1]]
#> NumericList of length 4
#> [["Channel"]] wAL00284RI5=-0.0371719605222012 ...
#> [["Set"]] wAL00284RI5=0.0994019567725992 ... wAL00286RI17=0.249998609880628
#> [["MedianIntensity"]] wAL00284RI5=0.119366566488542 ...
#> [["SampleType"]] wAL00284RI5=-0.382965748230049 ...
scpModelFilterThreshold(leduc_minimal)
#> [1] 1
####---- Setter ----####
scpModelFilterThreshold(leduc_minimal) <- 2
scpModelFilterThreshold(leduc_minimal)
#> [1] 2