Re-map data of Cufflinks paper by Trapnell et al.

[Tefina]

I am currently trying to work with the data from the Cufflinks paper by Trapnell et al. (http://www.nature.com/nbt/journal/v2.../nbt.1621.html).

E.g. the 60 hours sample contains 70 x 10^6 paired end reads, when I try to map the data with bowtie to the mouse genome, it only maps about 23%.
Of course, I would not expect that 100% map but intuitively I would assume that at least 50% should map.

Anybody also tried to remap this data? Or any other suggestions?

cheers,
tefina

[ShaunMahony]

It's RNA-seq data, and it's 75bp paired-end, so many reads span exon-exon junctions. Bowtie just tries to align to the genome -- it can't align spliced reads. Try aligning with Tophat instead

[ShaunMahony]

By the way, I've aligned this data using Tophat, and for the 60hr sample, I had ~21M "proper pairs", where both ends map in the correct orientation, and ~81M uniquely mapped ends (from ~70M x2)

[Tefina]

thanks a lot for sharing your experiences.

I have also thought that tophat probably makes more sense. I just wanted to try bowtie with the genomic sequence and see what I get..

[Tefina]

Any idea how to get the per base coverage per transcript from the SAM and the gtf file which is supplied in the GEO entry?