FindPeaks SeparateReads.jar error using GERALD/ELAND-generated files

[AdamB]

I'm trying to use FindPeaks with some data mapped in ELAND via GERALD. At the moment I'm stuck on using SeparateReads.jar. Below is a sample of the data from the input "s_X_sorted.txt" file:

HWI-EAS486	19	4	26	19288	17610	0	1	GCTTGTGTATCACTGGCGCAGCAGAGAGTCTGAGCATG	hhhhhhhhhghhhhhhfhfhcgdggcffchggfghfgh	chr10.fa		55	F	38131					
HWI-EAS486	19	4	65	12362	19698	0	1	CTGGTTAGAACAGCTAGGCCCTTGGGCCTCACTGTACA	hhhhhhghhhhhhhhhghhhhhhhhgfhhhhhhhfhhh	chr10.fa		145	F	38131					
HWI-EAS486	19	4	79	19730	5608	0	1	GCCAGCCTGGCAACTATCAGGCCACAAGGTGACATAAG	gggggggggdffc_fggggfc]ffWfcfadffffea_[	chr10.fa		309	F	19T18	88

I read in another thread that I could specify the aligner as "elandext", but to no avail. I get the following error message from SeparateReads.jar:

Version: Initializing class Log_Buffer                        $Revision: 2530 $
Version: Vancouver Short Read Analysis Package                4.0.16-dev.jars
Version: Initializing class SeparateReads                     $Revision: 3125 $
Version: Initializing class Generic_AlignRead_Iterator        $Revision: 3118 $
Version: Initializing class ElandExtIterator                  $Revision: 2933 $
Exception in thread "main" java.lang.StringIndexOutOfBoundsException: String index out of range: 38
	at java.lang.StringBuffer.setCharAt(StringBuffer.java:210)
	at src.lib.ioInterfaces.ElandExtUtilities.parse_mismatches(ElandExtUtilities.java:23)
	at src.lib.ioInterfaces.ElandExtIterator.next(ElandExtIterator.java:213)
	at src.lib.ioInterfaces.ElandExtIterator.next(ElandExtIterator.java:22)
	at src.lib.ioInterfaces.Generic_AlignRead_Iterator.hasNext(Generic_AlignRead_Iterator.java:121)
	at src.fileUtilities.SeparateReads.main(SeparateReads.java:70)

Is this a problem with the input file? I can't find any specification for the eland/elandext inputs for FindPeaks anywhere.

[apfejes]

Hi Adam,

This may be a little more technically challenging, but would you be willing to compile and run from source? I suspect that this problem has been solved in the latest version, which has not yet been released.

I can help walk you through the process, if you're unsure how to do that. If not, I can email you a compiled version.

Anthony

[AdamB]

Hi Anthony,

Yes I'd definitely like to try compiling the latest version. Is it on the website?

Thanks

[apfejes]

It's available from the svn repository. There are instructions here: http://sourceforge.net/apps/mediawik...=VSRAPBuilding

use svn to get a copy of the source, and ant to build it (once you have ant and subversion installed)

svn co http://vancouvershortr.svn.sourcefor...ershortr/trunk
ant buildall

[AdamB]

I can confirm the latest version (4.0.17) successfully separated my reads using the "elandext" flag.

Thanks for your help.

[apfejes]

Great, glad it's working!

Anthony

[AdamB]

Small update: I tried to generate a wig file, which worked apart from that the chromosome name in the header was "chr1.fa". UCSC did not like this, so I changed the chromosome name to "chr1", and then it loaded okay. Do you know where the ".fa" part is coming from, and if it can be changed?

Thanks.

[apfejes]

That's coming directly from your file - the aligner is using the name of the file as the chromosome. (You can even see it in the snippet you gave at the top of the thread for chr10.fa)

I seem to recall that there was a flag in FindPeaks to trim chromosome names and remove the .fa, but it escapes me what it would be. In the end, I think we just renamed all our .fa files to remove the extension and not have to worry about post processing our eland files.