New approach to computing correlations between pairs of overlapping features in terms of data matrices

E-P_pairs.R

library(plyranges)

set.seed(1)
x <- data.frame(seqnames=1, start=0:9 * 100 + 1, 
                width=20, id=1:10) %>%
  as_granges()

y <- data.frame(seqnames=1, start=round(runif(4,100,900)), 
                width=10, id=letters[1:4]) %>%
  as_granges() %>%
  sort()

tile <- data.frame(seqnames=1, start=c(1,401,801), 
                   end=c(400,800,1000), tile_id=1:3) %>%
  as_granges()

set.seed(1)
dat_x <- matrix(rnorm(10*100),nrow=10,dimnames=list(1:10,1:100))
dat_y <- matrix(rnorm(4*100),nrow=4,dimnames=list(letters[1:4],1:100))

x <- x %>% join_overlap_left(tile)
y <- y %>% join_overlap_left(tile)

x_overlaps <- x %>% 
  join_overlap_inner(y, maxgap=100) %>%
  filter(tile_id.x == tile_id.y) %>%
  select(tile_id = tile_id.x, id.x, id.y)

library(purrr)
x_overlaps %>%
  mutate(
    rho = map2_dbl(id.x, id.y, \(.x,.y) {
      cor(dat_x[.x,], dat_y[.y,])
    }))