qq_plot: Downsampled QQ-plot

In aaronwolen/gqq: Fast QQ-plots for GWAS results

Downsampled QQ-plot

Description

Downsampling reduces the number of points to be plotted. Since QQ-plots are on a logarithmic scale, near the origin (approaching 1) is very dense and most points are plotted over each other. However, the end of the distribution (approaching 0) is sparse. In order to make plotting more efficient by not plotting redundant points and also not lose information, there are two parameters needed for downsampling. The first is the proportion of points to randomly remove (-downsample). The second is the threshold (-pdown) for which points will be downsampled. For example, the options "-downsample 0.01 -pdown 10e-3" would random choose 1 p-values < 0.001 would be plotted. For one million tests, this reduces the number of points plotted to 10990 ((1,000,000*(1-0.001)*.01)+ (1,000,000*0.001)).

Usage

qq_plot(p, pdown = 3, downsample = 0.01, pch = 1, col = "grey20",
  cex = 0.1, qpoints = FALSE, ci.level = 0.9, ci.color = "grey70",
  highlight = NULL, cex.highlight = 1.2, max.axis = NULL)

Arguments

p	vector of p-values
pdown	p-value threshold (on a log10 scale) for plot downsampling
downsample	proportion of points to plot
pch	plotting 'character' (i.e., symbol) to use. See points for possible values.
col	color code or name, see par
cex	character (or symbol) expansion: a numerical vector. This works as a multiple of par("cex").
qpoints	logical, should the points be added to an existing plot?
ci.level	confidence level between 0 and 1. Set ci.level = NULL to avoid plotting confidence intervals (ignored if qpoints = TRUE)
ci.color	color of confidence interval band
highlight	p-values above this (log10) threshold will be plotted with points whose size is determined by cex.highlight, useful for highlighting strong associations of interest
cex.highlight	the magnification applied to points above the highlight threshold
max.axis	maximum range (on a log10 scale) of the X and Y axes

Examples

qq_plot(runif(1e5))

if (requireNamespace("pgcxd")) {
  data("scz", package = "pgcxd")
  qq_plot(scz$pval, highlight = 6)
}

aaronwolen/gqq documentation built on Feb. 17, 2024, 11:27 a.m.