coverageBed error: terminate called after throwing an instance of 'std::bad_alloc'

[hari_iyer16]

Hi all,
I am trying to run coverageBed from Bedtools in a cluster.
I connected to the cluster using the following parameters:
qrsh -l mem_free=10G,h_vmem=12G,h_stack=256M

Following is the command I ran:
coverageBed -a BED_Files/chip.bed -b hg18_chromInfo.txt | sortBed -i stdin > new/chip.coverage

where, hg18_chromInfo.txt is the file defining the bins/windows in which the sequencing coverage has to be summarized.

I get the following error:

terminate called after throwing an instance of 'std::bad_alloc'
what(): St9bad_alloc
/local/gcc-4.1.2/install/shortread/bin/coverageBed: line 2: 1221 Aborted ${0%/*}/bedtools coverage "$@"

I have no idea what the error means. Could someone please help troubleshooting. I tried googlins and seems that this error is not a specific instance of Bedtools error.

Thanks for your time.
Hari

[GenoMax]

Is the hg18_chromInfo.txt file in bed format?

[hari_iyer16]

@GenoMax: Yes, it is in Bed format. In fact the actual name used in the command is windows.bed. I changed the names while posting here to avoid cryptic names.

Here is the first few lines from the windows.bed or hg18_chromInfo.txt
chr1 0 199
chr1 200 399
chr1 400 599
chr1 600 799
chr1 800 999
chr1 1000 1199


Here is the first few lines from chip.bed
chr1 5 55 HWI-ST1191:621BA5ACXX:7:2306:6657:50696 1 -
chr1 14 62 HWI-ST1191:621BA5ACXX:6:1115:19691:10876 1 +
chr1 47 97 HWI-ST1191:621BA5ACXX:5:1310:19552:92789 1 +
chr1 47 97 HWI-ST1191:621BA5ACXX:5:1316:8676:35703 1 +
chr1 47 97 HWI-ST1191:621BA5ACXX:5:2316:17113:98272 1 +



Thanks.

[GenoMax]

Have you tried to capture both standard out and standard error from job submission?
Is the error you posted the only output?

How large are the two BED files you are using? Have these files been edited on Windows?

[hari_iyer16]

The chip.bed is 4.8G, and the windows.bed (hg18_chromInfo.txt) is 352M.
I figured the error is due to memory problems, and based on a suggestion from another mailing group, I used the -counts option in coverageBed:
coverageBed -counts -a BED_Files/chip.bed -b windows.bed | sortBed -i stdin > new/chip.coverage

However, still I am not able to run it because of memory issues. Got the following error by email [over the day at least I learnt to submit a batch job to the cluster and get a report of the job by email ;-) ]
Job 607506 (new) Aborted
Exit Status = 137
Signal = KILL
Start Time = 04/11/2013 21:50:39
End Time = 04/11/2013 22:01:39
CPU = 00:11:15
Max vmem = 12.024G
failed assumedly after job because:
job 607506.1 died through signal KILL (9)


I think I am not doing anything fancy here and many do this as routine. Generally for such a job, what memory specs are used?

Thanks.