plotOverlaps: plot cluster-cluster or segment-segment overlaps

In raim/segmenTools: Tools for Functional Genome Exploration

plot cluster-cluster or segment-segment overlaps

Description

Plots the significance distribution of cluster-cluster or segment-segment overlap statistics provided by clusterCluster, clusterProfile or segmentOverlaps, where a color gradient is calculated from -log(p), and the text shows the overlap numbers, e.g., the number of overlapping features for clusterCluster, or the Jaccard index or relative intersect values for segmentOverlaps. Option text allows to select which values to plot as text. Only "overlap" is available for a clusterCluster results, while for segmentOverlaps the Jaccard index ("jaccard"), or the relative intersect size can be shown: "intersect.target" and "intersect.query" are the intersect divided by total target or query length, respectively. Note that two distinct p-value cutoffs can be visualyized: p-values <=p.min are shown in black, and p-values p.txt are shown in white instead of black text (thus becoming visible on the black of significant overlaps).

Usage

plotOverlaps(
  x,
  p.min = 0.01,
  p.txt = p.min * 5,
  p.max,
  n = 100,
  col,
  values = c("overlap", "count", "statistic", "jaccard", "text"),
  type = 1,
  txt.col = c("black", "white"),
  rmz = TRUE,
  short = TRUE,
  scale = 1,
  round,
  axis = 1:2,
  show.sig = TRUE,
  show.total = FALSE,
  ...
)

Arguments

x	a ‘clusterOverlaps’ object returned by clusterCluster
p.min	significance cutoff, p-values equal or smaller to this cutoff will appear black (one-sided tests) or red/blue (two-sided tests)
p.txt	p-value cutoff for showing overlap numbers as white instead of black text
p.max	p-value cutoff for starting the color scale, e.g. 0.05; useful for generally high p-values.
n	number of color shades between p=1 (white) and p >= p.min (black)
col	color ramp, default are grey values (one-sided) or red (above) and blue (below) for two-sided tests, length of this vector overrules parameter n
values	selection of text (numeric values) to plot, depends on available data in x, see "Description"
type	1 for one-sided or 2 for two-sided tests color scheme, negative values of two-sided tests are reflected in negative p-values
txt.col	two colors used for the plot text, ie., the overlap counts; the second is used if 'p<p.txt' as a discrete signficance cutoff
rmz	remove 0 from text values
short	logical, indicating whether to cut higher overlap numbers; currently: division by 1000 and replacement by k
scale	factor to divide overlap numbers with, useful for low numbers in Jaccard index
round	number of digits to round overlap numbers to (useful for Jaccard index)
axis	integer vector, sets whether x-axis (1,3) and/or y-axis (2,4) are drawn; the column and row names of dat will be used as tick labels
show.sig	only for overlap lists sorted by sortClusters from package segmenTier: draws a red line where unsorted non-significant hits start
show.total	show total numbers (counts) of overlapping features on top and right axes
...	arguments to image_matrix

raim/segmenTools documentation built on March 20, 2024, 6:23 a.m.