Exclude short Reads?

[Greenhorn]

Hi,

I read some guidelines for RNA Seq and I am wondering why people discard reads under a certain threshold of bp and then only take reads into account which map unique. Even if no assembly or so is done.

Isn´t this somehow redundant? I mean, if a read has a length, which is sufficient to find a unique position, why should i discard it? I "only" want to map the reads to an annotaded genome to find DGE...

Thanks in advance

[dpryan]

If a read is too short then you don't bother mapping it. One doesn't normally map reads and then discard according to length for the reasons you noted.