Tophat: Is it necessary to pre-filter reads

[swarbre]

We have some short reads from the SRA which we plan to map with tophat is it necessary/standard to do any pre-filtering of the data prior to running tophat (i.e. removing reads with stretches of Ns or trimming poor quality 3' sequence).

Regards
dav

[Cole Trapnell]

I wouldn't do any kind of 3' trimming if it will leave your reads with differing lengths. TopHat expects that the reads will be the same length.

As far as filtering Ns, TopHat contains some very basic pre-alignment filtering to get rid of obviously failed or very low complexity reads, but if you want something more sophisticated than that, you'll need to do it yourself.