rankscore: compute rank^2 scores and rescale these between 0~1000 (with...

In goat: Gene Set Analysis Using the Gene Set Ordinal Association Test

compute rank^2 scores and rescale these between 0~1000 (with further precision captured by decimals)

Description

rank^2 can yield huge values, e.g. a large genelist of N=50000 genes would imply max gene score = 50000^2 = 2.5e+09. Thus we rescale these scores between 0~1000 so downstream applications (e.g. sum of 10000 gene scores) don't explode to huge numbers.

Usage

rankscore(x, sort1, sort2, sort3, colname)

Arguments

x	data.frame to sort (i.e. the genelist)
sort1	numeric vector of length nrow(x) to sort by first/primarily (descending order, higher value gets better result score)
sort2	numeric vector of length nrow(x) to sort by for breaking ties (descending order, higher value gets better result score)
sort3	numeric vector of length nrow(x) to sort by for breaking ties (descending order, higher value gets better result score)
colname	name for result column in x (overwritten if already exists)

Value

data.frame x with added column colname, containing gene scores between 0 and 1000

Examples

  x = data.frame(gene=c(1, 2, 3, 4, 5),
                 pvalue=c(0.01, 1, 1, 0.1, 1),
                 effectsize=c(-2, 0.25, 0.5, 1, 0.25))
  print(x, row.names = FALSE)
  print(rankscore(x, sort1 = -1*x$pvalue, sort2 = abs(x$effectsize),
        sort3 = x$gene, colname="score") |>
    arrange(desc(score)), row.names = FALSE)

goat documentation built on Aug. 27, 2025, 5:12 p.m.