75mer mapping with BOWTIE

[jwaage]

Hey all,

I'm looking for some input on tuning BOWTIE for 75mer RNA-seq reads from mus - any suggestions? I'm currently allowing only unique hits, and using 68bp as seed length, but that may be too conservative.

Best,
Johannes Waage
Uni. of Copenhagen

[Ben Langmead]

Hi Johannes,

That's certainly reasonable; I suggest doing a series of runs using your target data but also using the -u parameter to cut the job off after a certain number of reads so that it doesn't take forever. Doing several such runs with different sets of parameters (but the same -u) should help you determine the sweet spot w/r/t speed and sensitivity.

Ben

[graveley]

Hi,

We are doing this routinely. We typically use the -v 2 option and use an index made from both the genome and splice junctions with 69 nt on either side of the junction. Surprisingly, we get a very large fraction of reads that map uniquely using these parameters.

Brent

[xuying]

Hi Johannes:
How do you deal with exon junctions? some exons will be shorter than 75bps. Try tophat?

Xuying

[drio]

Can you tell me what reference are you using?

[jwaage]

Can you tell me what reference are you using?

mm9 combined with a combinatorial exon-exon DB (142mers, requiring overlap of 4nt)