memory allocation error running deadzones

[Li-Wei]

Hi,

I am going to use rseg for my ChIP-seq dataset, which has a tag size
of 41bp. The rseg program website does not have deadzone files for
hg19 and read length 41 so I tried to generate it with the deadzone
program. The author has mentioned that earlier he used ~11gb memory
and 17hr to finish the deadzone run for mm9. However, I assigned 16Gb
memory but the program still terminated with a "could not allocate
memory" error. Has anyone ever run deadzones successfully using hg19?
Would it be okay if I run rseg without a deadzone file?

Thanks!
Li-Wei

[Song, Qiang]

Hi Li-Wei,

First of all, using deadzone correction significantly improves the
performance of RSEG.  It is worth the effort to compute deadzones.

For your convenience, we precomputed the deadzone file of 41mer in
hg19. It is downloadable from
http://smithlab.cmb.usc.edu/histone/rseg/data/deadzones-k41-hg19.bed.bz2

When you are using the deadzone program, you may change the -prefix
option to adjust memory usage. The larger this option is, the more
memory the program consumes and the faster the program runs. The
default value is 5. If deadzone cannot get enough memory, try a
smaller -prefix value.

Regards,
Song Qiang