Hi there,

A little late after the fact, I just noticed the "new" tileGenome function in GenomicRange - very nice. Thank you!  I'll start using this now instead of a much slower function I'd written myself. 

I sometimes find myself looking at overlapping sliding windows with my slow function, rather than the non-overlapping tiles that tileGenome produces. Would it be possible to add that option to the function?  I'd like to specify window size and slide amount. In a slightly ridiculous toy example, I might want 100bp windows with a slide of 20bp on a 150bp chromosome, so the windows would be at these positions:

1-100

21-120

41-140

61-150

What do you think - would this be easy for you guys to do?

thanks very much,

Janet

Perhaps you can get what you want with available operations.

>    seqlengths <- c(chr1=150, chr2=500)

> tt = tileGenome(seqlengths, tilewidth=20)

> utt = unlist(tt)

> trim(resize(utt, width=100))

GRanges object with 34 ranges and 0 metadata columns:

       seqnames     ranges strand

          <Rle>  <IRanges>  <Rle>

   [1]     chr1  [ 1, 100]      *

   [2]     chr1  [21, 120]      *

   [3]     chr1  [41, 140]      *

   [4]     chr1  [61, 150]      *

   [5]     chr1  [80, 150]      *

   ...      ...        ...    ...

  [30]     chr2 [403, 500]      *

  [31]     chr2 [423, 500]      *

  [32]     chr2 [442, 500]      *

  [33]     chr2 [462, 500]      *

  [34]     chr2 [482, 500]      *

  -------

  seqinfo: 2 sequences from an unspecified genome

Thanks very much, all - it's nice to have a way to do it with existing code, and also great to see that it could be a built-in option for the function at some point soon.  The built-in option will be much more intuitive for us naive biologists than the clever resizing method, which isn't immediately obvious.

Herve: yes, that idea does sound very reasonable.  Perhaps it'd help people search for and intuitively understand the new option if the help page includes the phrases "sliding window" and "overlap", even if the option is called spacing - I think those might be the more commonly used names in publications, etc. 

thanks again,

Janet

Thanks, Herve - that's great!

Janet

Just want to continue to illustrate the approach @Martin showed. The 6th range, which is at the end of the following snippet, may not be desirable. It is completely within the second last range.
 

> seqlengths <- c(chr1=60, chr2=20, chr3=25)
> tt <- tileGenome(seqlengths, tilewidth = 20, cut.last.tile.in.chrom = TRUE)
> tt
GRanges object with 6 ranges and 0 metadata columns:
      seqnames    ranges strand
         <Rle> <IRanges>  <Rle>
  [1]     chr1  [ 1, 20]      *
  [2]     chr1  [21, 40]      *
  [3]     chr1  [41, 60]      *
  [4]     chr2  [ 1, 20]      *
  [5]     chr3  [ 1, 20]      *
  [6]     chr3  [21, 25]      *
  -------
  seqinfo: 3 sequences from an unspecified genome
>seqlengths(tt)[as.character(seqnames(tt))]
chr1 chr1 chr1 chr2 chr3 chr3
  60   60   60   20   25   25
> seqnames(tt)
factor-Rle of length 6 with 3 runs
  Lengths:    3    1    2
  Values : chr1 chr2 chr3
Levels(3): chr1 chr2 chr3
> as.character(seqnames(tt))
[1] "chr1" "chr1" "chr1" "chr2" "chr3" "chr3"
>  
>
> end(tt) <- pmin(start(tt) + 30, seqlengths(tt)[as.character(seqnames(tt))])
> tt
GRanges object with 6 ranges and 0 metadata columns:
      seqnames    ranges strand
         <Rle> <IRanges>  <Rle>
  [1]     chr1  [ 1, 31]      *
  [2]     chr1  [21, 51]      *
  [3]     chr1  [41, 60]      *
  [4]     chr2  [ 1, 20]      *
  [5]     chr3  [ 1, 25]      *
  [6]     chr3  [21, 25]      * # Should this range be here??!
  -------
  seqinfo: 3 sequences from an unspecified genome
>