BASiCS_TestDE: Detection of genes with changes in expression

In catavallejos/BASiCS: Bayesian Analysis of Single-Cell Sequencing data

Detection of genes with changes in expression

Description

Function to assess changes in expression between two groups of cells (mean and over-dispersion)

Usage

BASiCS_TestDE(
  Chain1,
  Chain2,
  EpsilonM = log2(1.5),
  EpsilonD = log2(1.5),
  EpsilonR = log2(1.5)/log2(exp(1)),
  ProbThresholdM = 2/3,
  ProbThresholdD = 2/3,
  ProbThresholdR = 2/3,
  OrderVariable = c("GeneIndex", "GeneName", "Mu"),
  GroupLabel1 = "Group1",
  GroupLabel2 = "Group2",
  Plot = TRUE,
  PlotOffset = TRUE,
  PlotOffsetType = c("offset estimate", "before-after", "MA plot"),
  Offset = TRUE,
  EFDR_M = 0.05,
  EFDR_D = 0.05,
  EFDR_R = 0.05,
  GenesSelect = rep(TRUE, ncol(Chain1@parameters[["mu"]])),
  min.mean = 1,
  MinESS = 100,
  ...
)

Arguments

Chain1	an object of class BASiCS_Chain containing parameter estimates for the first group of cells
Chain2	an object of class BASiCS_Chain containing parameter estimates for the second group of cells
EpsilonM	Minimum fold change tolerance threshold for detecting changes in overall expression (must be a positive real number). Default value: EpsilonM = log2(1.5) (i.e. 50% increase).
EpsilonD	Minimum fold change tolerance threshold for detecting changes in biological over-dispersion (must be a positive real number). Default value: EpsilonM = log2(1.5) (i.e. 50% increase).
EpsilonR	Minimum distance threshold for detecting changes in residual over-dispersion (must be a positive real number). Default value: EpsilonR= log2(1.5)/log2(exp(1)) (i.e. 50% increase).
ProbThresholdM	Optional parameter. Probability threshold for detecting changes in overall expression (must be a positive value, between 0 and 1). If EFDR_M = NULL, the posterior probability threshold for the differential mean expression test will be set to ProbThresholdM. If a value for EFDR_M is provided, the posterior probability threshold is chosen to achieve an EFDR equal to EFDR_M and ProbThresholdM defines a minimum probability threshold for this calibration (this avoids low values of ProbThresholdM to be chosen by the EFDR calibration. Default value ProbThresholdM = 2/3, i.e. the probability of observing a log2-FC above EpsilonM must be at least twice the probality of observing the complementary event (log2-FC below EpsilonM).
ProbThresholdD	Optional parameter. Probability threshold for detecting changes in cell-to-cell biological over-dispersion (must be a positive value, between 0 and 1). Same usage as ProbThresholdM, depending on the value provided for EFDR_D. Default value ProbThresholdD = 2/3.
ProbThresholdR	Optional parameter. Probability threshold for detecting changes in residual over-dispersion (must be a positive value, between 0 and 1). Same usage as ProbThresholdM, depending on the value provided for EFDR_R. Default value ProbThresholdR = 2/3.
OrderVariable	Ordering variable for output. Possible values: 'GeneIndex' (default), 'GeneName' and 'Mu' (mean expression).
GroupLabel1	Label assigned to reference group. Default: GroupLabel1 = 'Group1'
GroupLabel2	Label assigned to reference group. Default: GroupLabel2 = 'Group2'
Plot	If Plot = TRUE, MA and volcano plots are generated.
PlotOffset	If Plot = TRUE, the offset effect is visualised.
PlotOffsetType	See argument Type in BASiCS_PlotOffset for more information.
Offset	Optional argument to remove a fix offset effect (if not previously removed from the MCMC chains). Default: Offset = TRUE.
EFDR_M	Target for expected false discovery rate related to the comparison of means. If EFDR_M = NULL, EFDR calibration is not performed and the posterior probability threshold is set equal to ProbThresholdM. Default EFDR_M = 0.05.
EFDR_D	Target for expected false discovery rate related to the comparison of dispersions. If EFDR_D = NULL, EFDR calibration is not performed and the posterior probability threshold is set equal to ProbThresholdD.Default EFDR_D = 0.05.
EFDR_R	Target for expected false discovery rate related to the comparison of residual over-dispersions. If EFDR_R = NULL, EFDR calibration is not performed and the posterior probability threshold is set equal to ProbThresholdR.Default EFDR_D = 0.05.
GenesSelect	Optional argument to provide a user-defined list of genes to be considered for the comparison. Default: GenesSelect = rep(TRUE, nGene) When used, this argument must be a vector of TRUE (include gene) / FALSE (exclude gene) indicator, with the same length as the number of intrinsic genes and following the same order as how genes are displayed in the table of counts. This argument is necessary in order to have a meaningful EFDR calibration when the user decides to exclude some genes from the comparison.
min.mean	Minimum mean expression threshold required for inclusion in offset calculation. Similar to 'min.mean' in 'scran::computeSumFactors'. This parameter is only relevant with 'Offset = TRUE'.
MinESS	The minimum effective sample size for a gene to be included in the tests for differential expression. This helps to remove genes with poor mixing from differential expression tests. Default is 100. If set to NA, genes are not checked for effective sample size before differential expression tests are performed.
...	Optional parameters.

Value

BASiCS_TestDE returns an object of class BASiCS_ResultsDE

Author(s)

Catalina A. Vallejos [e-mail]

Nils Eling [e-mail]

Examples

# Loading two 'BASiCS_Chain' objects (obtained using 'BASiCS_MCMC')
data(ChainSC)
data(ChainRNA)

Test <- BASiCS_TestDE(
  Chain1 = ChainSC, Chain2 = ChainRNA,
  GroupLabel1 = "SC", GroupLabel2 = "P&S",
  EpsilonM = log2(1.5), EpsilonD = log2(1.5),
  OffSet = TRUE
)

# Results for the differential mean test
head(as.data.frame(Test, Parameter = "Mean"))

# Results for the differential over-dispersion test
# This only includes genes marked as 'NoDiff' in Test$TableMean
head(as.data.frame(Test, Parameter = "Disp"))

# For testing differences in residual over-dispersion, two chains obtained
# via 'BASiCS_MCMC(Data, N, Thin, Burn, Regression=TRUE)' need to be provided
data(ChainSCReg)
data(ChainRNAReg)

Test <- BASiCS_TestDE(
  Chain1 = ChainSCReg, Chain2 = ChainRNAReg,
  GroupLabel1 = 'SC', GroupLabel2 = 'P&S',
  EpsilonM = log2(1.5), EpsilonD = log2(1.5),
  EpsilonR = log2(1.5)/log2(exp(1)),
  OffSet = TRUE
)

## Plotting the results of these tests
BASiCS_PlotDE(Test)

catavallejos/BASiCS documentation built on March 27, 2024, 12:49 a.m.