combineBlocks: Combine blockwise statistics

In MarioniLab/scran: Methods for Single-Cell RNA-Seq Data Analysis

Combine blockwise statistics

Description

Combine DataFrames of statistics computed separately for each block. This usually refers to feature-level statistics and sample-level blocks.

Usage

combineBlocks(
  blocks,
  ave.fields,
  pval.field,
  method,
  geometric,
  equiweight,
  weights,
  valid
)

Arguments

blocks	A list of DataFrames containing blockwise statistics. These should have the same number of rows and the same set of columns.
ave.fields	Character vector specifying the columns of blocks to be averaged. The value of each column is averaged across blocks, potentially in a weighted manner.
pval.field	String specifying the column of blocks containing the p-value. This is combined using combineParallelPValues.
method	String specifying how p-values should be combined, see ?combineParallelPValues.
geometric	Logical scalar indicating whether the geometric mean should be computed when averaging ave.fields.
equiweight	Logical scalar indicating whether each block should be given equal weight.
weights	Numeric vector of length equal to blocks, containing the weight for each block. Only used if equiweight=TRUE.
valid	Logical vector indicating whether each block is valid. Invalid blocks are still stored in the per.block output but are not used to compute the combined statistics.

Value

A DataFrame containing all fields in ave.fields and the p-values, where each column is created by combining the corresponding block-specific columns. A per.block column is also reported, containing a DataFrame of the DataFrames of blockwise statistics.

Author(s)

Aaron Lun

See Also

This function is used in modelGeneVar and friends, combineVar and testLinearModel.

Examples

library(scuttle)
sce <- mockSCE()

y1 <- sce[,1:100] 
y1 <- logNormCounts(y1) # normalize separately after subsetting.
results1 <- modelGeneVar(y1)

y2 <- sce[,1:100 + 100] 
y2 <- logNormCounts(y2) # normalize separately after subsetting.
results2 <- modelGeneVar(y2)

# A manual implementation of combineVar:
combineBlocks(list(results1, results2), 
    ave.fields=c("mean", "total", "bio", "tech"),
    pval.field='p.value', 
    method='fisher',
    geometric=FALSE,
    equiweight=TRUE,
    weights=NULL,
    valid=c(TRUE, TRUE))

MarioniLab/scran documentation built on Sept. 7, 2024, 6:25 a.m.