map subsetByOverlaps back to input coordinates

Question

[brtawe]

I am trying to find overlap between my own set of genomic coordinates and genes, as shown below. I am able to get get the subset of genes that overlap using subsetByOverlaps, however, how do I map the genes back to the coordinates that I used to subset? I want to know which of my input coordinates overlapped with each gene(s)

I included a metadata column in my coordinates containing an ID, but that ID is missing from the subset output

mycoords.gr = as.data.frame(input) %>%
 makeGRangesFromDataFrame(keep.extra.columns=TRUE) #make GRanges form Bed file

genes <- genes(TxDb.Hsapiens.UCSC.hg38.knownGene) #extract GRanges for genes
gene_list <- subsetByOverlaps(genes, mycoords.gr) #subset genes that overlap with my coordinates

gene_result = as(gene_list, "data.frame") #convert GRanges object to data frame

Thanks for any help!

Answer 1

[rpolicastro]

Example data.

library("GenomicFeatures")

my_coords <- structure(list(seqnames = structure(c(1L, 1L, 1L, 1L), class = "factor", .Label = "I"), 
    start = c(5L, 10L, 15L, 20L), end = c(8L, 13L, 18L, 23L), 
    width = c(4L, 4L, 4L, 4L), strand = structure(c(1L, 2L, 1L, 
    2L), class = "factor", .Label = c("+", "-", "*"))), class = "data.frame", row.names = c(NA, 
-4L))
my_coords <- makeGRangesFromDataFrame(my_coords)

> my_coords
GRanges object with 4 ranges and 0 metadata columns:
      seqnames    ranges strand
         <Rle> <IRanges>  <Rle>
  [1]        I       5-8      +
  [2]        I     10-13      -
  [3]        I     15-18      +
  [4]        I     20-23      -
  -------
  seqinfo: 1 sequence from an unspecified genome; no seqlengths

genes <- structure(list(seqnames = c("I", "I"), start = c(1, 21), end = c(9, 
30), strand = c("+", "-"), name = c("ENSG006069", "ENSG006241"
)), class = "data.frame", row.names = c(NA, -2L))
genes <- makeGRangesFromDataFrame(genes, keep.extra.columns=TRUE)

> genes
GRanges object with 2 ranges and 1 metadata column:
      seqnames    ranges strand |        name
         <Rle> <IRanges>  <Rle> | <character>
  [1]        I       1-9      + |  ENSG006069
  [2]        I     21-30      - |  ENSG006241

I find plyranges::join_overlap_left_directed is simpler than the base GenomicRanges method for this task.

library("plyranges")

overlaps <-  join_overlap_left_directed(my_coords, genes)

> overlaps
GRanges object with 4 ranges and 1 metadata column:
      seqnames    ranges strand |        name
         <Rle> <IRanges>  <Rle> | <character>
  [1]        I       5-8      + |  ENSG006069
  [2]        I     10-13      - |        <NA>
  [3]        I     15-18      + |        <NA>
  [4]        I     20-23      - |  ENSG006241

[brtawe]

That works great, thanks for the help!