geneSetAssoc

Perform self-contained or competitive gene set analyses.

geneSetAssoc(
  object,
  geneSet = NULL,
  scoreMatrix = NULL,
  cormatrix = NULL,
  condition = NULL,
  covar = NULL,
  test = c("lm", "mlm", "fisher", "ttest", "ztest", "ACAT"),
  threshold = NULL,
  Zcutoffs = NULL,
  INT = FALSE,
  scoreCutoffs = NULL,
  minSetSize = 1L,
  maxSetSize = Inf,
  oneSided = TRUE,
  memlimit = 1000L,
  ID = "unit",
  output = NULL,
  verbose = TRUE
)

Arguments

object: An rvbResult object.
geneSet: A geneSetList or geneSetFile object.
scoreMatrix: A matrix (rows = genes, columns = features) can be provided to perform enrichment analyses on continuous values. These can be used to perform e.g. cell-type enrichment analyses.
cormatrix: a correlation matrix with row and column names corresponding to the units in the rvbResult. Needs to be specified in order to run the 'mlm' (mixed linear model) test. A burden score correlation matrix can be generated using the buildCorMatrix method. The mixed linear model is run using the GENESIS R package.
condition: Perform conditional analyses. Input can be 1) a geneSetList or geneSetFile, in which case the genesets specified in the geneSet parameter will all be conditioned on the gene sets provided here. 2) a vector of gene set names present in geneSet, all genesets specified in the geneSetList/geneSetFile will be conditioned on the genesets specified here.
covar: A vector of covariates
test: Vector of tests to perform. Currently implemented tests are the competitive tests lm,mlm,fisher and self-contained tests ttest,ztest and ACAT.
threshold: A vector of thresholds for cutoff-based tests (fisher's exact test).
Zcutoffs: A vector (length=2, minimum and maximum) of cutoffs to apply to the Z-scores. Z scores below/above these cutoffs will be set equal to the cutoff.
INT: Apply inverse normal transformation to Z-scores? Defaults to FALSE.
scoreCutoffs: If scoreMatrix is specified, this parameter can be set to cap scores in the scorematrix. It should be a vector of length 2 (minimum and maximum sd). Defaults to NULL, in which case no score cutoffs are applied.
minSetSize: Exclude genesets with size < minSetSize. Defaults to 1.
maxSetSize: Exclude genesets with size > maxSetSize. Defaults to Inf.
oneSided: Calculate a one-sided P-value? Defaults to TRUE.
memlimit: Maximum number of genesets to process in one go.
ID: ID column in the rvbResult that corresponds with the IDs used in the geneSetList. Defaults to 'unit'.
output: Optional: save results to specified file path.
verbose: Should the function be verbose? Defaults to TRUE.

Value

An object of class gsaResult or a data.frame if scoreMatrix is used. The columns are described below:

geneSetName: Name of the gene set.
test: Statistical test used.
covar: Covariates included.
threshold: Threshold used for cutoff-based tests.
geneSetSize: Number of genes in the gene set.
genesObs: Number of genes observed in the input results.
effect: Effect size estimate.
effectSE: Standard error of the effect size estimate.
effectCIlower: Lower confidence interval of the effect size estimate.
effectCIupper: Upper confidence interval of the effect size estimate.
P: P-value.

References

Gogarten SM, Sofer T, Chen H, Yu C, Brody JA, Thornton TA, Rice KM, Conomos MP. Genetic association testing using the GENESIS R/Bioconductor package. Bioinformatics. 2019 Dec 15;35(24):5346-5348

Examples

library(rvatData)
data(rvbresults)
res <- rvbresults[
  rvbresults$test == "firth" &
    rvbresults$varSetName == "ModerateImpact",
]

# example genesetlist used in examples below (see ?buildGeneSet on build geneSetLists/geneSetFiles)
genesetlist <- buildGeneSet(
  list(
    "geneset1" = c("SOD1", "NEK1"),
    "geneset2" = c("ABCA4", "SOD1", "NEK1"),
    "geneset3" = c("FUS", "NEK1")
  )
)

# Perform competitive gene set analysis using a linear model
GSAresults <- geneSetAssoc(
  res,
  genesetlist,
  covar = c("nvar"),
  test = c("lm")
)
#> 1 Z-scores are +Inf, these are set to the maximum observed Z-score: 4.219.
#> 3 out of 3 sets are kept.

# Outlying gene association scores can be remedied by either setting Z-score cutoffs (i.e. all Z-scores exceeding these values will be set to the respective cutoff),
# or inverse normal transforming the Z-scores:
GSAresults <- geneSetAssoc(
  res,
  genesetlist,
  covar = c("nvar"),
  test = c("lm"),
  maxSetSize = 500,
  Zcutoffs = c(-4, 4) # lower and upper bounds
)
#> 0 Z-scores <-4 are set to -4
#> 3 Z-scores >4 are set to 4
#> 3 out of 3 sets are kept.

GSAresults <- geneSetAssoc(
  res,
  genesetlist,
  covar = c("nvar"),
  test = c("lm"),
  maxSetSize = 500,
  INT = TRUE # perform inverse normal transformation
)
#> 3 out of 3 sets are kept.


# Conditional gene set analyses can be performed to test whether gene sets are associated independently with the phenotype of interest.
# In the example below we test whether gene sets are independent of geneset1
GSAresults <- geneSetAssoc(
  res,
  condition = getGeneSet(genesetlist, "geneset1"),
  geneSet = genesetlist,
  covar = c("nvar"),
  test = c("lm"),
  maxSetSize = 500
)
#> 1 Z-scores are +Inf, these are set to the maximum observed Z-score: 4.219.
#> 3 out of 3 sets are kept.

# perform two-sided tests by setting `oneSided = FALSE`
GSAresults <- geneSetAssoc(
  res,
  genesetlist,
  covar = c("nvar"),
  test = c("lm"),
  maxSetSize = 500,
  oneSided = TRUE
)
#> 1 Z-scores are +Inf, these are set to the maximum observed Z-score: 4.219.
#> 3 out of 3 sets are kept.

# Test whether the proportion of P-values below a specified threshold is greater than the proportion outside of it.
# Using Fisher's exact test
# The `threshold` parameter specifies the P-value cutoff to define significant genes:
GSAresults <- geneSetAssoc(
  res,
  genesetlist,
  test = c("fisher"),
  threshold = 1e-4,
  maxSetSize = 500
)
#> 1 Z-scores are +Inf, these are set to the maximum observed Z-score: 4.219.
#> 3 out of 3 sets are kept.

Arguments

Value

References

See also

Examples