results: Extracting results and aberrant splicing events

In c-mertes/FRASER: Find RAre Splicing Events in RNA-Seq Data

Extracting results and aberrant splicing events

Description

The result function extracts the results from the given analysis object based on the given options and cutoffs. The aberrant function extracts aberrant splicing events based on the given cutoffs.

Usage

## S4 method for signature 'FraserDataSet'
results(
  object,
  sampleIDs = samples(object),
  padjCutoff = 0.1,
  deltaPsiCutoff = 0.1,
  rhoCutoff = NA,
  aggregate = FALSE,
  collapse = FALSE,
  minCount = 5,
  psiType = psiTypes,
  geneColumn = "hgnc_symbol",
  all = FALSE,
  returnTranscriptomewideResults = TRUE,
  additionalColumns = NULL,
  BPPARAM = bpparam()
)

## S4 method for signature 'FraserDataSet'
aberrant(
  object,
  type = fitMetrics(object),
  padjCutoff = 0.1,
  deltaPsiCutoff = 0.1,
  minCount = 5,
  rhoCutoff = NA,
  by = c("none", "sample", "feature"),
  aggregate = FALSE,
  geneColumn = "hgnc_symbol",
  subsetName = NULL,
  all = FALSE,
  ...
)

Arguments

object	A FraserDataSet object
sampleIDs	A vector of sample IDs for which results should be retrieved
padjCutoff	The FDR cutoff to be applied or NA if not requested.
deltaPsiCutoff	The cutoff on delta psi or NA if not requested.
rhoCutoff	The cutoff value on the fitted rho value (overdispersion parameter of the betabinomial) above which junctions are filtered
aggregate	If TRUE the returned object is aggregated to the feature level (i.e. gene level).
collapse	Only takes effect if aggregate=TRUE. If TRUE, collapses results across the different psi types to return only one row per feature (gene) and sample.
minCount	The minimum count value of the total coverage of an intron to be considered as significant. result
psiType	The psi types for which the results should be retrieved.
geneColumn	The column name of the column that has the gene annotation that will be used for gene-level pvalue computation.
all	By default FALSE, only significant introns (or genes) are listed in the results. If TRUE, results are assembled for all samples and introns/genes regardless of significance.
returnTranscriptomewideResults	If FDR corrected pvalues for subsets of genes of interest have been calculated, this parameter indicates whether additionally the transcriptome-wide results should be returned as well (default), or whether only results for those subsets should be retrieved.
additionalColumns	Character vector containing the names of additional columns from mcols(fds) that should appear in the result table (e.g. ensembl_gene_id). Default is NULL, so no additional columns are included.
BPPARAM	The BiocParallel parameter.
type	Splicing type (psi5, psi3 or theta)
by	By default none which means no grouping. But if sample or feature is specified the sum by sample or feature is returned
subsetName	The name of a subset of genes of interest for which FDR corrected pvalues were previously computed. Those FDR values on the subset will then be used to determine aberrant status. Default is NULL (using transcriptome-wide FDR corrected pvalues).
...	Further arguments can be passed to the method. If "n", "padjVals", "dPsi" or "rhoVals" are given, the values of those arguments are used to define the aberrant events.

Value

For results: GRanges object containing significant results. For aberrant: Either a of logical values of size introns/genes x samples if "by" is NA or a vector with the number of aberrant events per sample or feature depending on the vaule of "by"

Examples

# get data, fit and compute p-values and z-scores
fds <- createTestFraserDataSet()

# extract results: for this example dataset, no cutoffs are used to
# show the output of the results function
res <- results(fds, all=TRUE)
res

# aggregate the results by genes (gene symbols need to be annotated first 
# using annotateRanges() function)
results(fds, padjCutoff=NA, deltaPsiCutoff=0.1, aggregate=TRUE)

# aggregate the results by genes and collapse over all psi types to obtain 
# only one row per gene in the results table 
results(fds, padjCutoff=NA, deltaPsiCutoff=0.1, aggregate=TRUE, 
        collapse=TRUE)

# get aberrant events per sample: on the example data, nothing is aberrant
# based on the adjusted p-value
aberrant(fds, type="jaccard", by="sample")

# get aberrant events per gene (first annotate gene symbols)
fds <- annotateRangesWithTxDb(fds)
aberrant(fds, type="jaccard", by="feature", padjCutoff=NA, aggregate=TRUE)
        
# find aberrant junctions/splice sites
aberrant(fds, type="jaccard")

# retrieve results limiting FDR correction to only a subset of genes
# first, we need to create a list of genes per sample that will be tested
geneList <- list('sample1'=c("TIMMDC1"), 'sample2'=c("MCOLN1"))
fds <- calculatePadjValues(fds, type="jaccard", 
                 subsets=list("exampleSubset"=geneList))
results(fds, all=TRUE, returnTranscriptomewideResults=FALSE)

c-mertes/FRASER documentation built on June 14, 2024, 7:49 p.m.