plot_disco: Create a disco plot

In bihealth/bioshmods: What the Package Does (One Line, Title Case)

Create a disco plot

Description

Make a concordance / discordance plots for two contrasts

Usage

plot_disco(
  contrast1,
  contrast2,
  lower = -100,
  upper = 100,
  show_top_labels = 0,
  top_labels_both = TRUE,
  annot1 = NULL,
  annot2 = NULL,
  alpha = 0.5,
  disco = NULL,
  by = 0,
  primary_id = "PrimaryID",
  label_col = "SYMBOL",
  label_sel = NULL
)

Arguments

contrast1, contrast2	data frames with rownames corresponding to IDs (they don't need to be in the same order) and columns log2FoldChange and pvalue.
lower, upper	lower and upper boundaries for coloring of the score
show_top_labels	sort the genes by descending absolute disco score and show top N labels
top_labels_both	should top labels from both negative and positive disco scores be shown, or only for the absolute top, whether only negative or only positive or both?
annot1, annot2	annotation data frames for the two contrasts
alpha	transparency
disco	result of disco_score; if provided, it will be used to avoid unnecessary computations
by	column by which the contrast data frames should be merged (passed to merge). Default: merge by row names
primary_id	the name which should be assigned to the identifier column which results from the merge
label_col	name of the column in annot which should be used for labels
label_sel	identifiers of labels to be shown on the plot

Details

A concordance / discordance plot is simply a log~2~ FC vs log~2~ FC comparison between two contrasts. Disco score is a heuristic score reflecting the strength of similarity or dissimilarity between two log~2~ FC values weighted by the corresponding p-values and given by the formula log2FC.x * log2FC.y * (-log10(pval.x) - log10(pval.y)).

To manually select which labels to show, use a truncated annot object – remove all rows you don't want to show, and set the show_top_labels parameter to Inf.

Value

a ggplot object (plot)

Examples

## Generate example data
c1 <- data.frame(log2FoldChange=rnorm(5000, sd=2))
c1$pvalue <- pnorm(abs(c1$log2FoldChange), sd=2, lower.tail=FALSE)
c1$PrimaryID <- paste0("ID", 1:nrow(c1))

c2 <- data.frame(log2FoldChange=c1$log2FoldChange + rnorm(5000, sd=3))
c2$pvalue <- pnorm(abs(c2$log2FoldChange), sd=2, lower.tail=FALSE)
c2$PrimaryID <- paste0("ID", 1:nrow(c2))

## Example disco plot
plot_disco(c1, c2)

bihealth/bioshmods documentation built on July 1, 2023, 4:32 a.m.