Batman ears in our read distributions

[Mchicken]

Hey guys,

like everyone else we are observing uneven/non-uniform coverage in our RNA-Seq experiments and I know that there are several things influencing this coverage issue like GC-bias or PCR-bias.

But what we also observe are two dominant peaks located towards both (5 and 3 prime) ends of a transcript (we call them "batman ears").

In the paper of Roberts et al. (http://www.genomebiology.com/2011/12/3/R22/) they mention the following:

Positional bias refers to a local effect in which fragments are preferentially located towards either the beginning or end of transcripts

This might fit to our observations although we observe it for both the beginning and end of transcripts.

Do you guys have any idea what is the reason for such positional bias?

Thanks
Mario