dist_desc: Return discriptive statistics about effect sizes/p-values.

In hemstrow/GeneArchEst: Estimate Genomic Architecture for Quantitative Traits

Return discriptive statistics about effect sizes/p-values.

Description

Calculates a range of descriptive statistics both about the p-values as a whole, effect size peaks, and sliding windows.

Usage

dist_desc(p, meta, windows, delta, pcut = 0.005, chr = "chr", pvals = T)

Arguments

p	numeric. p-values or effect sizes for each SNP.
delta	numeric. Value used to determine spacing between peaks.
pcut	numeric, default 0.005. Only p-values/effect sizes above/below this quantile will be used for peak detection.
chr	character,default "chr". Name of the meta column containing chromosome info.
pvals	logical, default TRUE. Specifies if p-values are provided. p-values will be -log10 transformed.
meta.	data.frame. The position and chromosome for each SNP. Needs a column named "position".

Details

Peaks are identified with a peak finding algorithm based on the delta and pcut values. These values should be chosen based on the true p-value or effect size distribution. p-values are -log10 transformed.

Stats about windows are run through the basic stat calculation twice: once per window (means per window, for example), then once *per stat* (so the mean of the mean values per window).

The following stats are calculated:

• mean

• median

• sd

• variance

• 20 evenly spaced quantiles between 0 and 1

• kurtosis

• skewness

.

In addition, the number of unique peaks is also calculated.

hemstrow/GeneArchEst documentation built on June 10, 2025, 5:06 a.m.