[BioC] rtracklayer GraphTrackLine

[Dario Strbenac]

Yeah, that'd be a cool feature, since we're not doing any data manipulations with it - just getting a set of screenshots.

Thanks for all the help over the last couple days !

---- Original message ----
>Date: Tue, 11 May 2010 05:44:40 -0700
>From: Michael Lawrence <lawrence.michael at gene.com>  
>Subject: Re: [BioC] rtracklayer GraphTrackLine  
>To: D.Strbenac at garvan.org.au
>Cc: Michael Lawrence <lawrence.michael at gene.com>, bioconductor at stat.math.ethz.ch
>
>   On Tue, May 11, 2010 at 1:18 AM, Dario Strbenac
>   <D.Strbenac at garvan.org.au> wrote:
>
>     Thanks, I'm happy to use either way.
>
>     Just 1 more question, I now tried to do a simple
>     example with just 1 bigWig because it was running
>     slowly, and my example is :
>
>     library(rtracklayer)
>     library(grDevices)
>     session <- browserSession("UCSC")
>     track1 <-
>     import("http://129.94.136.7/bw/Bre12p3_MBD2IP_1.bw",
>     selection = GenomicSelection("hg18", colnames =
>     "score")) # Using ver. 1.8.1 still.
>     track1 <- as(track1, "UCSCData")
>     track(session, "Bre12 P3", priority = 1,
>     viewLimits = c(0, 10), color =
>     as(col2rgb("green"), "integer")) <- track1
>
>     The actual .bw is only 438 Mb, so I'm curious
>     about the resource (memory) usage ?
>
>      PID USER      PR  NI  VIRT  RES  SHR S
>     %CPU %MEM    TIME+  COMMAND
>     10572 darstr    20   0 9375m 9.1g 4876 S    0
>     57.7  15:26.58 R
>
>     It's also curious that processing has stopped at
>     this stage (i.e. I didn't get the > prompt yet),
>     because the hard drive is hardly swapping, and I
>     can open another R session that goes at 100% CPU
>     (it's run on an 8 core computer).
>
>   Sounds to me like it's network overhead. If the
>   bigWig is 438 MB, then the wig/bedGraph data is
>   going to be huge. It's important to realize that
>   rtracklayer does not send tracks to UCSC as bigWig.
>   This is because you need to give UCSC a URL on a web
>   server. All rtracklayer has is a RangedData object
>   in memory. Would it be useful for you to have
>   rtracklayer write it out to some specified location,
>   with a specified URL to pass to UCSC? In your case,
>   you really only need to pass the URL.
>
>   Michael


--------------------------------------
Dario Strbenac
Research Assistant
Cancer Epigenetics
Garvan Institute of Medical Research
Darlinghurst NSW 2010
Australia