Produce a weitrix of shift scores — counts

Produce a weitrix of shift scores between -1 and 1. The input is read counts at a collection of peaks (or other features) in a collection of samples. The peaks can be grouped by gene, and are ordered within each gene.

counts_shift(counts, grouping, verbose = TRUE, typecast = identity)

Arguments

counts	A matrix of read counts. Rows are peaks and columns are samples.
grouping	A data frame defining the grouping of peaks into genes. Should have a column "group" naming the gene and a column "name" naming the peak (corresponding to `rownames(counts)`). Within each group, peak names should be ordered from 5' to 3' position.
verbose	If TRUE, output some debugging and progress information.
typecast	A function to convert a matrix to a matrix-like type. Applied at the chunk level, before all chunks are `rbind`ed. Allows use of memory-efficient matrix representations.

Value

A SummarizedExperiment object with metadata fields marking it as a weitrix.

Details

For a particular gene, a shift score measures measures the tendency of reads to be upstrand (negative) or downstrand (positive) of the average over all samples. Shift scores range between -1 and 1.

Examples

grouping <- data.frame(
    group=c("A","A","A","B","B"),
    name=c("p1","p2","p3","p4","p5"))

counts <- rbind(
    p1=c(1,2,0),
    p2=c(0,1,0),
    p3=c(1,0,0),
    p4=c(0,0,1),
    p5=c(0,2,1))

wei <- counts_shift(counts, grouping)
#> Calculating shifts in 1 blocks

weitrix_x(wei)
#>   [,1]       [,2]  [,3]
#> A  0.2 -0.1333333    NA
#> B   NA  0.2500000 -0.25
weitrix_weights(wei)
#>   [,1] [,2] [,3]
#> A    2    3    0
#> B    0    2    2
rowData(wei)
#> DataFrame with 2 rows and 2 columns
#>   per_read_var total_reads
#>      <numeric>   <numeric>
#> A       0.2560           5
#> B       0.1875           4