combineBlocks: Combine blockwise statistics
In scran: Methods for Single-Cell RNA-Seq Data Analysis

Description Usage Arguments Value Author(s) See Also Examples

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

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

`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 `combinePValues`.
`method`	String specifying how p-values should be combined, see `?combinePValues`.
`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.

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.

Aaron Lun

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

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))