In metamaden/recountmethylationManuscriptSupplement: Supplement for the manuscript "Human methylome variation across Infinium 450K data on the Gene Expression Omnibus"

library(recountmethylationManuscriptSupplement)
pkgname <- "recountmethylationManuscriptSupplement"
scripts.dir <- system.file("scripts", "figures", 
                           package = pkgname)

library(ggplot2)
library(gridExtra)
library(data.table)

knitr::opts_chunk$set(echo = TRUE, eval = FALSE)

Figure 1

Some details about the platforms of interest are as follows:

lid <- list(c('GPL13534','HM450K','2011','May 2011'),
            c('GPL21145', 'EPIC','2015','Nov. 2015'),
            c('GPL8490', 'HM27K','2009','Apr. 2009')
)
df <- as.data.frame(do.call(rbind, lid))
colnames(df) <- c("platform_id", "platform_name", 
                  "release_year", "release_month")
knitr::kable(df)

1A Available GEO samples (GSM IDs) by year, platform, and type

Obtain sample (GSM IDs) and study (GSE IDs) summaries from queries to GEO with queries to Entrez utilities. With Entrez utilities installed, navigate to the directory with the script "eqplot.py" and run:

python3 eqplot.py

This will cause data queries to the GEO servers, and result in the creation of several new files, "gsmyeardata", "gsmidatyrdata", "gseyeardata", and "gseidatyrdata". Finally, run the script "fig1a.R" to make the plot.

script.name <- "fig1a.R"
source(file.path(scripts.dir, script.name))
fig1a

1B

See data analyses vignette in recountmethylation

Figure 2

2A BeadArray control outcomes

Source the script "fig2a.R". This creates control outcomes from BeadArray signals in Table S2, then generates the plot object.

script.name <- "fig2a.R"
source(file.path(scripts.dir, script.name))
fig2a

2B Quality signals, by storage condition

See data analyses vignette in recountmethylation

2C Percent BeadArray control failure, by storage condition

script.name <- "fig2c.R"
source(file.path(scripts.dir, script.name))
fig2c

2D Study-level quality assessments

script.name <- "fig2d.R"
source(file.path(scripts.dir, script.name))
draw(hmlist, row_title = paste0("Study ID (N = ", 
                                length(gsesize.subset),")"), 
     column_title = "Sub-threshold Frequency", 
     heatmap_legend_side = "right", annotation_legend_side = "right")

Figure 3 PCA of autosomal DNAm

Details including sample identification, analyses, and results summaries are detailed in the script pca_fig3.R. This script generates the files contained in r system.file("extdata", "pcadata", package = pkgname), which are used below to generate the main plots.

3A

Run the following to generate the scatterplot of the top two components from PCA of all samples. Red points are noncancer blood, purple points are leukemia samples.

script.name <- "fig3a.R"
source(file.path(scripts.dir, script.name))
fig3a

3B

Run the following to generate the scatterplot of the top two components from PCA of all samples except blood or leukemias. Blue points are noncancer brain, dark cyan points are brain tumors.

script.name <- "fig3b.R"
source(file.path(scripts.dir, script.name))
fig3b

3C

Run the following to generate the scatterplot of the top two components from PCA of 7 noncancer tissues: sperm (blue); adipose (dark red); blood (red); brain (purple); buccal (orange); nasal (light green); and liver (dark green).

script.name <- "fig3c.R"
source(file.path(scripts.dir, script.name))
fig3c

3D

Run the following to generate the scatterplot of the top two components from PCA of 6 noncancer tissues: adipose (dark red); blood (red); brain (purple); buccal (orange); nasal (light green); and liver (dark green).

script.name <- "fig3d.R"
source(file.path(scripts.dir, script.name))
fig3d

Figure 4 DNAm variability in noncancer tissues

For analysis details, see also the data analyses vignette in recountmethylation

4A

Violin plots of 14,000 high variance probes

script.name <- "fig4ab.R"
source(file.path(scripts.dir, script.name))
fig4a

fig4b

4B

This section shows how to generate the composite image of stacked barplots showing probe genome region mappings among the 14,000 probes with high tissue-specific variances in 7 noncancer tissues.

script.name <- "fig4c.R"
suppressMessages(source(file.path(scripts.dir, script.name)))

metamaden/recountmethylationManuscriptSupplement documentation built on May 31, 2021, 5:27 a.m.