In perishky/meffil: Efficient algorithms for DNA methylation

EWAS report

• study: r study
• author: r author
• date: r format(Sys.time(), '%d %B, %Y')

Parameters

ewas.summary$parameters

Sample characteristics

For continuous or ordinal variables, the "mean" column provides the mean and the "var" column the standard deviation of the variable. For categorical variables, the "mean" column provides the number of samples with the given "value" and the "var" column the percentage of samples with the given "value".

knitr::kable(ewas.summary$sample.characteristics)

Covariate associations

out <- NULL
if (!is.null(ewas.summary$covariate.associations)) {
    for (covariate.name in names(ewas.summary$covariate.associations))
        out <- c(out, knit_child(file.path(report.path, "ewas-covariate.rmd")))
}

if (!is.null(out))
    cat(out, sep="\n\n")

QQ plots

out <- NULL
for (plot in ewas.summary$qq.plots)
    out <- c(out, knit_child(file.path(report.path, "qq-plot.rmd")))

cat(out, sep="\n")

Manhattan plots

out <- NULL
for (plot in ewas.summary$manhattan.plots)
    out <- c(out, knit_child(file.path(report.path, "manhattan-plot.rmd")))

cat(out, sep="\n")

Significant CpG sites

There were r length(ewas.summary$significant.sites) CpG sites with association p-values < r ewas.summary$parameters$sig.threshold. These are listed in the file associations.csv.

tab <- with(ewas.summary, cpg.stats[match(significant.sites, rownames(cpg.stats)),])
write.csv(tab, file=file.path(opts_knit$get("output.dir"), "associations.csv"))

The following table shows overlaps between associations under different sets of covariates:

p.value.idx <- grep("p.value", colnames(tab))
tab.overlaps <- apply(tab[,p.value.idx,drop=F], 2, function(p.a) {
    apply(tab[,p.value.idx,drop=F], 2, function(p.b) {
        sum(p.a < ewas.summary$parameters$sig.threshold
            & p.b < ewas.summary$parameters$sig.threshold, na.rm=T)
    })
})
knitr::kable(tab.overlaps)

practical.sites <- names(ewas.summary$cpg.plots)
tab <- with(ewas.summary, cpg.stats[match(practical.sites, rownames(cpg.stats)),])

Below are the r length(practical.sites) CpG sites with association p-values < r ewas.summary$parameters$practical.threshold in the r ewas.summary$parameters$model regression model.

knitr::kable(tab[practical.sites,])

Plots of these sites follow, one for each covariate set. "p[lm]" denotes the p-value obtained using a linear model and "p[beta]" the p-value obtained using beta regression.

out <- NULL
for (cpg in practical.sites) 
    out <- c(out, knit_child(file.path(report.path, "cpg-plot.rmd")))

cat(out, sep="\n")

Selected CpG sites

Number of CpG sites selected: r length(ewas.summary$selected.sites).

tab <- with(ewas.summary, cpg.stats[match(selected.sites, rownames(cpg.stats)),])
knitr::kable(tab)

out <- NULL
for (cpg in ewas.summary$selected.sites)
    out <- c(out, knit_child(file.path(report.path, "cpg-plot.rmd")))

cat(out, sep="\n\n")

R session information

sessionInfo()

perishky/meffil documentation built on June 9, 2024, 5:59 p.m.