Illumina RNA-seq: should I trim low quality bases prior to mapping?

[JonB]

I have Illumina sequences from a cDNA sample. I want to map the reads to a reference genome and calculate expression levels. I think I will map using bowtie/tophat.

The adaptors are removed from my sequences, but I wonder whether I need to trim away the low quality bases at the 3'-end of the reads? Or is this taken into account in the mapping?

Thanks!

[pbluescript]

Tophat can sometimes have trouble mapping reads with a lot of low quality bases on the end. Trimming the reads might help with that, but shorter reads are more likely to map to multiple locations. If they are paired end reads, you will need to reassess your mean inner distance. Try it with and without the trimming and see which works best for your data.
STAR is another alignment program you might consider trying that will automatically soft clip bases off a read if necessary to find a good alignment.