connect_adjacent_clusters: Connect adjacent clusters when their sources are within a...

In cnobles/gintools: Genomic DNA Integration Analysis Tools

Description

Given a list of unique integration site positions (reduced GRanges object) and a directed graph of connected components, this function identifies clusters where source nodes are within a gap distance away from each other, and connects the source nodes with a directional edge, based first on abundance, and secondly on an upstream bias for tie breaking.

Usage

connect_adjacent_clusters(red.sites, graph, gap, bais)

Arguments

red.sites	GRanges object which has been reduced to single nt positions and contains the revmap from the original GRanges object. The object must also contain a column for cluster membership (clusID) and a column for abundance (fragLengths).
graph	a directed graph built from the red.sites object. Each node corresponds to a row in the red.sites object.
bias	either "upsteam" or "downstream", designating which position to choose if other decision metrics are tied.

Details

connect_adjacent_clusters returns a graph where adjacent clusters with sources within the gap distance are joined.

Author(s)

Christopher Nobles, Ph.D.

Examples

gr <- gintools:::generate_test_granges(stdev = 3)
red.sites <- reduce(
  flank(gr, -1, start = TRUE),
  min.gapwidth = 0L,
  with.revmap = TRUE)
red.sites$siteID <- seq_along(red.sites)
revmap <- as.list(red.sites$revmap)
red.sites$fragLengths <- lengths(revmap)
red.hits <- GenomicRanges::as.data.frame(
  findOverlaps(red.sites, maxgap = 0L, drop.self = TRUE))
red.hits <- red.hits %>%
  mutate(q_pos = start(red.sites[queryHits])) %>%
  mutate(s_pos = start(red.sites[subjectHits])) %>%
  mutate(q_abund = red.sites[queryHits]$abundance) %>%
  mutate(s_abund = red.sites[subjectHits]$abundance) %>%
  mutate(strand = unique(strand(
    c(red.sites[queryHits], red.sites[subjectHits])))) %>%
  mutate(is.upstream = ifelse(
    strand == "+",
    q_pos < s_pos,
    q_pos > s_pos)) %>%
  mutate(keep = q_abund > s_abund) %>%
  mutate(keep = ifelse(
    q_abund == s_abund,
    is.upstream,
    keep)) %>%
  filter(keep)
g <- make_empty_graph(n = length(red.sites), directed = TRUE) %>%
  add_edges(unlist(mapply(
    c, red.hits$queryHits, red.hits$subjectHits, SIMPLIFY = FALSE)))
red.sites$clusID <- clusters(g)$membership
g <- connect_satalite_vertices(red.sites, g, gap = 2L, "upstream")
red.sites$clusID <- clusters(g)$membership
g <- break_connecting_source_paths(red.sites, g, "upstream")
red.sites$clusID <- clusters(g)$membership

connect_adjacent_clusters(red.sites, g, gap = 5L, "upstream")

cnobles/gintools documentation built on Aug. 22, 2019, 10:36 a.m.