breakpoint error

[TimL]

Hi all:

I'm trying to run a multi-SNP analysis (just using 2-SNP mapping) on a
large SNP dataset (≈300 million markers). The single SNP mapping runs
work great, but when I try to do a 2-SNP mapping run, I get the
following error (I've included the command line as well):

Belk217BiMac:~ tlightfoot$ /Applications/BimBam/./bimbam_mac -g /
Projects/HAM_epistatic_project/data/Genomic-desnesnp.txt -p /Projects/
HAM_epistatic_project/data/38_grp_HAM_dist.csv -pos /Projects/
HAM_epistatic_project/data/Genomic-location.txt -o epistatic_dist -l
2
-bimbam: position files contain 8234637 records.
-bimbam: file 0 has 38 individual and 8234636 snps
-bimbam: exclude 8443 snps due to maf < 0
-bimbam: read file 0 again
-bimbam: single snp analysis of typed snps:
[>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>] 100%
bimbam_mac(189) malloc: *** mmap(size=1082724144119808) failed (error
code=12)
*** error: can't allocate region
*** set a breakpoint in malloc_error_break to debug
terminate called after throwing an instance of 'std::bad_alloc'
what(): St9bad_alloc
Abort trap
Belk217BiMac:~ tlightfoot$

Any ideas?

Tim

[Bertrand Servin]

Hi,

I might be wrong but from the message :

,bimbam_mac(189) malloc: *** mmap(size=1082724144119808) failed (error
> code=12)

I would say Bimbam is trying to allocate a 10^15 bytes array, that's a
lot (about 1 Petabyte, yes that's a 1000 Terabytes) and you probably
don't have that much memory available on your machine.

--
Bertrand

[Matthew Stephens]

yes, trying to do all pairs of markers among 300 million is rather ambitious.
This functionality was designed to do all pairs of markers among, say,
a few hundred.

Matthew