getter_setter_functions {FRASER} | R Documentation |
This is a collection of small accessor/setter functions for easy access to the values within the FRASER model.
featureExclusionMask(fds, type = currentType(fds)) featureExclusionMask(fds, type = currentType(fds)) <- value rho(fds, type = currentType(fds)) zScores(fds, type = currentType(fds), byGroup = FALSE, ...) pVals(fds, type = currentType(fds), level = "site", dist = "BetaBinomial", ...) padjVals(fds, type = currentType(fds), dist = c("BetaBinomial"), ...) predictedMeans(fds, type = currentType(fds)) deltaPsiValue(fds, type = currentType(fds)) currentType(fds) currentType(fds) <- value pseudocount(value = NULL) hyperParams(fds, type = currentType(fds), all = FALSE) bestQ(fds, type = currentType(fds)) dontWriteHDF5(fds) dontWriteHDF5(fds) <- value verbose(fds) verbose(fds) <- value
fds |
An FraserDataSet object. |
type |
The type of psi (psi5, psi3 or theta) |
value |
The new value to be assigned. |
byGroup |
If TRUE, aggregation by donor/acceptor site will be done. |
... |
Internally used parameteres. |
level |
Indicates if the retrieved p values should be adjusted on the donor/acceptor site-level (default) or if unadjusted junction-level p values should be returned. |
dist |
Distribution for which the p-values should be extracted. |
all |
Logical value indicating whether |
A (delayed) matrix or vector dependent on the type of data retrieved.
featureExclusionMask
: Retrieves a logical vector indicating
for each junction whether it is included or excluded during the fitting
procedure.
featureExclusionMask<-
: To remove certain junctions from
being used in the train step of the encoding dimension we can set the
featureExclusion
vector to FALSE
. This can be helpfull
if we have local linkage between features which we do not want to
model by the autoencoder.
rho
: Returns the fitted rho values for the
beta-binomial distribution
zScores
: This returns the calculated z-scores.
pVals
: This returns the calculated p-values.
padjVals
: This returns the adjusted p-values.
predictedMeans
: This returns the fitted mu (i.e. psi)
values.
deltaPsiValue
: Returns the difference between the
observed and the fitted psi values.
currentType
: Returns the psi type that is used
within several methods in the FRASER package.
currentType<-
: Sets the psi type that is to be used
within several methods in the FRASER package.
pseudocount
: Sets and returns the pseudo count used
within the FRASER fitting procedure.
hyperParams
: This returns the results of the
hyperparameter optimization NULL if the hyperparameter
opimization was not run yet.
bestQ
: This returns the optimal size of the
latent space according to the hyperparameter optimization or a simple
estimate of about a tenth of the number of samples if the hyperparameter
opimization was not run yet.
dontWriteHDF5
: Gets the current value of whether the
assays should be stored as hdf5 files.
dontWriteHDF5<-
: Sets whether the assays should be stored
as hdf5 files.
verbose
: Dependend on the level of verbosity
the algorithm reports more or less to the user. 0 means being quiet
and 10 means everything.
verbose<-
: Sets the verbosity level to a value
between 0 and 10. 0 means being quiet and 10 means reporting everything.
fds <- createTestFraserDataSet() # should assays be saved as hdf5? dontWriteHDF5(fds) dontWriteHDF5 <- TRUE # get/set the splice metric for which results should be retrieved currentType(fds) <- "psi5" currentType(fds) # get fitted parameters bestQ(fds) predictedMeans(fds) rho(fds) # get statistics pVals(fds) padjVals(fds) zScores(fds) # set and get pseudocount pseudocount(4L) pseudocount() # retrieve or set a mask to exclude certain junctions in the fitting step featureExclusionMask(fds, type="theta") <- sample( c(FALSE, TRUE), nrow(mcols(fds, type="theta")), replace=TRUE) featureExclusionMask(fds, type="theta") # controlling the verbosity level of the output of some algorithms verbose(fds) <- 2 verbose(fds)