On Fri, Oct 1, 2010 at 6:58 AM, Kasper Daniel Hansen <
[email protected]> wrote:

> Seems like the required changes hasn't yet been pushed from svn to the
> builds.
>
> I do have one possible caveat: it is not clear to me how this would
> work when the GRanges is not sorted by chromosome as I get
>
> > gr1
> GRanges with 3 ranges and 0 elementMetadata values
>    seqnames    ranges strand |
>       <Rle> <IRanges>  <Rle> |
> [1]     chr1    [1, 3]      * |
> [2]     chr2    [2, 4]      * |
> [3]     chr1    [6, 8]      * |
>
> seqlengths
>  chr1 chr2
>   NA   NA
> > as(gr1, "RangesList")
> CompressedIRangesList of length 2
> $chr1
> IRanges of length 2
>    start end width
> [1]     1   3     3
> [2]     6   8     3
>
> $chr2
> IRanges of length 1
>    start end width
> [1]     2   4     3
>
> When I go back to the original GRanges I will need to interweave the
> results from the RangesList.
>
>
Yes, that's one issue. Easiest to just sort the GRanges up-front. Btw,
RangedData can store things in interval trees without a separate object.



> Of course, being able to do IntervalTrees directly on GRanges will
> also be very convenient.
>


I agree. But that would take some work, since IntervalTrees have no notion
of chromosome.

Kasper
>
>
> On Fri, Oct 1, 2010 at 12:05 AM, Michael Lawrence
> <[email protected]> wrote:
> > One easy way to get something like this:
> >
> > rangesList <- seqapply(as(gr, "RangesList"), IntervalTree)
> >
> > Then you'll have a RangesList of IntervalTree objects for range queries.
> >
> > Might need latest IRanges svn for this to work reliably (problem with
> > subsetting IRanges with metadata).
> >
> > On Thu, Sep 30, 2010 at 6:48 PM, Kasper Daniel Hansen
> > <[email protected]> wrote:
> >>
> >> I would like to see 'IntervalTrees for GRanges'.  Specifically, if I
> >> am repeatedly doing findOverlaps lookups on a big GRanges, I want to
> >> be able to store the IntervalTrees that (right now?) gets computed
> >> each time.
> >>
> >> This may of course already exist, in which case I would be satisfied
> >> with a pointer :)
> >>
> >> Thanks,
> >> Kasper
> >>
> >> _______________________________________________
> >> Bioc-sig-sequencing mailing list
> >> [email protected]
> >> https://stat.ethz.ch/mailman/listinfo/bioc-sig-sequencing
> >
> >
>

        [[alternative HTML version deleted]]

_______________________________________________
Bioc-sig-sequencing mailing list
[email protected]
https://stat.ethz.ch/mailman/listinfo/bioc-sig-sequencing

Reply via email to