simulate_null.HiCglm: Samples interactions based on background model.

Description Usage Arguments Value Examples

Description

Based on this method output one may plot expected vs observed plot of interactions for each diagonal pool (see examples).

Usage

1
2
## S3 method for class 'HiCglm'
simulate_null(hic.glm, dpools = "all", N = 10)

Arguments

hic.glm

HiCglm object

dpools

character or vector, if character "all" then use all pooled diagonals in background model, otherwise use specified subset

N

integer number of samples for every predictor value

Value

list of list of data frames with expected interactions, first list enumerates different maps (usually chromosomes), second list contains 2 elements refering to Y | X and X | Y models. Finally the last 2 lists contain data frames with simulated interactions.

Examples

1
2
3
4
5
6
7
8
# first prepare HiCglm object
expected <- simulate_null(hicglm, dpools = 1:10)
observed <- lapply(hicglm$model[["18"]][as.character(1:10)], function(x) x$data)
observed <- do.call("rbind", observed)[c("val.x", "val.y", "dpool")]
# plot results: expected vs observed
ggplot(all.data, aes(x = val.x, y = val.y, color = type)) +
  geom_point(size = 0.3, alpha = 0.7) +
  facet_wrap(~ dpool, scales = "free", ncol = 3)

rz6/DIADEM documentation built on Dec. 31, 2019, 3:51 a.m.