Significant boundaries and read density

[Andreas Gschwind]

Hi,

I am using rseg to call various histone modifications such as
h3k27me3, h3k4me3 and h3k4me1. When visually comparing enriched
domains with enriched domains that have significant boundaries
associated with them in regard to read density, I noticed some strange
results. I often observe enriched domains that show nice enrichment,
relatively clear boundaries and high boundary scores but do not have
significant boundaries, while others that seem worse are associated
with significant boundaries. Especially in the case of h3k4me1 it's
very extreme, where most domains with significant boundaries are very
short and show low enrichment, while the larger domains with high
enrichment mostly don't have significant boundaries. I used the
default parameters for the -posterior-cutoff and -cdf-cutoff options.
Did you experience similar situations or have any idea what might
cause this?

Thanks a lot!

regards, Andreas