Strange tophat2 results

Question

[mbio.kyle]

I am working with some viral RNA sequenceing data. I have attempted to use tophat to align the reads and have noticed two things.

1. The run times are exceedingly long, I have a resonably powerful workstation (dual 6 core processors, 32GB RAM and a RAID array) and tophat just takes forever. 10X longer then bowtie. I imagine this perhaps to be expected?
2. The second, and more concerning is the results. I have run the samples through tophat2 and through bowtie2 to compare the two and see what is happening. Here is an image of the results in wig format.
   1. image link

I know that tophat is supposed to be mapping across splice junctions. Is that what I am seeing? It seems to me like it is finding alot more then bowtie2, so much so that i am worried. I know this is a very vauge question. I appreciate anyones opinons or two cents on this

Thanks, Kyle

Answer 1

[Brian Bushnell]

The image seems to indicate coverage across introns; assuming that deletion locations are being treated as "covered", that's what it would look like. I prefer IGV; it makes it much more obvious what's happening.

As far as splice-aware aligners go, Tophat is very slow; BBMap is substantially faster and more sensitive. However, note that alignments across splices do inherently require more computational effort than unspliced alignments.