Optimizing settings for mapping 100bp RNA-seq reads with tophat

[seqhorn]

Any ideas for optimizing tophat settings when mapping long reads (100bp RNA-seq single) to a reference genome?

So far we map around 30% of the reads using default settings, which is much lower than we have commonly observed.

Previously we were able to map more than 70% of 50bp reads to the same reference genome. We have also tried further quality filtering without raising the mapping rate.

Maybe another tool is better for long reads, but we would like to run cufflinks on the tophat output for differential expression/splicing.

Any thoughts?

[seqhorn]

I have managed to increase the mapping rate to 70% by running bowtie separately and raising the values for -n and -e options. Is there a way to have tophat parse the same settings on to bowtie?

[frozenlyse]

Have you checked for adaptor contamination in your reads?

[seqhorn]

Thanks for replying, i think we finally identified the problem, seems like something went wrong during library construction. So really not a mapping issue.