Example code for joining two range sets where second one is also attached to an SE

join_se.R

library(SummarizedExperiment)
library(plyranges)

# example data
m <- matrix(rnorm(600), nrow=100)
r1 <- data.frame(seqnames=1, start=1:50 * 100 + 2501,
                 width=5, id1=paste0("u",formatC(1:50,width=3,flag="0"))) |>
  as_granges()
r2 <- data.frame(seqnames=1, start=1:100 * 100 + 1,
                 width=5, id2=paste0("v",formatC(1:100,width=3,flag="0"))) |>
  as_granges()
rownames(m) <- r2$id2
colnames(m) <- paste0("sample",1:6)
# this matrix is associated wit r2 (and regions id2)
se <- SummarizedExperiment(list(counts=m),
                           rowRanges=r2,
                           colData=DataFrame(z=1:6))

# the function should do someting like this:

# join based on ranges
olap <- r1 |>
  join_overlap_inner(r2, maxgap=100)

# expand the SE
olap_se <- se[olap$id2,] # contains duplicate rows sometimes
rowData(olap_se)$id1 <- olap$id1 # add IDs from the query ranges
rowData(olap_se)$combined <- with(rowData(olap_se), paste(id1, id2, sep="-"))

library(dplyr)
# compute something, e.g. pick the range with top variance in the assay data
idx <- olap_se |>
  rowData() |>
  as_tibble() |>
  mutate(variance = rowVars(assay(olap_se))) |>
  group_by(id1) |>  
  slice(which.max(variance)) |>
  ungroup() |>
  arrange(id1)

# subset to an SE that now matches the size of query ranges
final_se <- olap_se[ match(idx$combined, rowData(olap_se)$combined), ]

join_se_rows.R

library(SummarizedExperiment)
library(plyranges)

# example data
m <- matrix(rnorm(600), nrow=100)
r1 <- data.frame(seqnames=1, start=1:50 * 100 + 2501,
                 width=5, id1=paste0("u",formatC(1:50,width=3,flag="0"))) |>
  as_granges()
r2 <- data.frame(seqnames=1, start=1:100 * 100 + 1,
                 width=5, id2=paste0("v",formatC(1:100,width=3,flag="0"))) |>
  as_granges()
rownames(m) <- r2$id2
colnames(m) <- paste0("sample",1:6)
# this matrix is associated wit r2 (and regions id2)
se <- SummarizedExperiment(list(counts=m),
                           rowRanges=r2,
                           colData=DataFrame(z=1:6))

# the function should do someting like this:

# join based on ranges
olap <- r1 |>
  join_overlap_inner(r2, maxgap=100)

# expand the SE
olap_se <- se[olap$id2,] # contains duplicate rows sometimes
rowData(olap_se)$id1 <- olap$id1 # add IDs from the query ranges
rowData(olap_se)$combined <- with(rowData(olap_se), paste(id1, id2, sep="-"))

library(dplyr)
library(tidyr)
library(purrr)

rowData(olap_se) |>
  as_tibble() |>
  select(id1) |>
  bind_cols(assay(olap_se)) |>
  nest(data = -id1) |>
  mutate(row_ave = map(data, \(mat) {
    colMeans(mat)
  })) |>
  pull(row_ave) |>
  bind_rows() |>
  as.matrix()