[Bioc-sig-seq] Rolexa run-time error in batch mode

Seth Falcon sfalcon at fhcrc.org
Fri Nov 13 00:19:48 CET 2009


Hi Michael,

Sorry I don't have more, but a few hints are below...

On 11/12/09 2:48 PM, Michael Muratet wrote:
> which grinds along for awhile and then I get a "subscript out of bounds"
> error in a method with which I am unfamiliar. See traceback:
>
>  > traceback()
> 18: stop("subscript out of bounds")
> 17: .nextMethod(x = x, i = i, ...)
> 16: .Call("R_nextMethodCall", call, callEnv, PACKAGE = "methods")
> 15: callNextMethod()
> 14: x at ranges[i]
> 13: x at ranges[i]
> 12: seqInit at sread[colonies]
> 11: seqInit at sread[colonies]
> 10: .local(run, ...)
> 9: SeqScore(run = run, int = int, seqInit = seq, colonies = 1:dim(int)[1],
> cycles = 1:ncycles)
> 8: SeqScore(run = run, int = int, seqInit = seq, colonies = 1:dim(int)[1],
> cycles = 1:ncycles)
> 7: .local(run, ...)
> 6: OneBatch(run, path, lane, tiles[fn + 1:nfiles], outpath, prefix)
> 5: OneBatch(run, path, lane, tiles[fn + 1:nfiles], outpath, prefix)
> 4: ForkBatch(run = rolenv.1, sp.1, outpath = "./", prefix = "rs.1",
> nthreads = 2, nfiles = 2, lane = lane, tiles = 1:2)
> 3: eval.with.vis(expr, envir, enclos)
> 2: eval.with.vis(ei, envir)
> 1: source("batch_analysis.R")
>
> I haven't been able to locate the method where the error occurs. Maybe
> it's in R itself somewhere. Can anyone point me in the right direction?

Frames 11, 12 are from the Rolexa package, I believe.  Line 213 of 
Rolexa/R/Model.R has:

                       siseq = seqInit at sread[colonies]

Then frames 13, 14 jump to either IRanges or Biostrings code (or both).

Have you been able to run a perhaps smaller example without the fork 
package?

I'm not familiar with fork, but assuming that it is forking the R 
process, one might be suspicious of using fork with packages that use 
external pointers such as Biostrings.  Since R is not multi-threaded, 
package code is not setup for concurrent access... Or this might have 
nothing to do with the issue you are seeing :-P

+ seth

-- 
Seth Falcon
Program in Computational Biology | Fred Hutchinson Cancer Research Center



More information about the Bioc-sig-sequencing mailing list