> On Apr 5, 2016, at 10:27 AM, 何尧 <he...@pku.edu.cn> wrote:
>
> I do have a bunch of genes ( nearly ~50000) from the whole genome, which
> read in genomic ranges
>
> A range(gene) can be seem as an observation has three columns chromosome,
> start and end, like that
>
> seqnames start end width strand
>
> gene1 chr1 1 5 5 +
>
> gene2 chr1 10 15 6 +
>
> gene3 chr1 12 17 6 +
>
> gene4 chr1 20 25 6 +
>
> gene5 chr1 30 40 11 +
>
> I just wondering is there an efficient way to find overlapped, upstream and
> downstream genes for each gene in the granges
The data.table package (in CRAN) and the iRanges package (in bioC) have
formalized efficient approaches to those problems.
>
> For example, assuming all_genes_gr is a ~50000 genes genomic range, the
> result I want like belows:
>
> gene_nameupstream_genedownstream_geneoverlapped_gene
> gene1NAgene2NA
> gene2gene1gene4gene3
> gene3gene1gene4gene2
> gene4gene3gene5NA
>
> Currently , the strategy I use is like that,
> library(GenomicRanges)
> find_overlapped_gene <- function(idx, all_genes_gr) {
> #cat(idx, "\n")
> curr_gene <- all_genes_gr[idx]
> other_genes <- all_genes_gr[-idx]
> n <- countOverlaps(curr_gene, other_genes)
> gene <- subsetByOverlaps(curr_gene, other_genes)
> return(list(n, gene))
> }
>
> system.time(lapply(1:100, function(idx) find_overlapped_gene(idx,
> all_genes_gr)))
> However, for 100 genes, it use nearly ~8s by system.time().That means if I
> had 50000 genes, nearly one hour for just find overlapped gene.
>
> I am just wondering any algorithm or strategy to do that efficiently, perhaps
> 50000 genes in ~10min or even less
>
I suspect this would happen on a much faster basis for such a small dataset.
--
David.
> [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
David Winsemius
Alameda, CA, USA
______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
