TopHat questions (indexes and seq length)

[statsteam]

Hi all,

I just downloaded and installed tophat with bowtie.

Using the test_data, it worked fine and found some junctions.

I went ahead and ran Chris Burge's cerebellum data (12M reads, 36bp) on h_sapiens_37_asm (hg19). It ran about 2.5 hours and to my surprise there is no junctions.

Should I just use h_sapiens (hg18) instead of h_sapiens_37_asm?
What are the difference between assembly and contig indexes?

I know that the tophat chops the seqeunce into small pieces and tries to map the small ones. Is there a way we can specify the length of small pieces? I think, in my run, it did not chop the initial sequence (36bp).

Thank you,
Statsteam

[Cole Trapnell]

This is unfortunately a known bug in TopHat v1.0.10, caused by the fact that the h_sapiens_37_asm index contains spaces in the index record names, and TopHat parses the Bowtie output and the adjoining fasta files for the index inconsistently. The workaround is to index hg19 yourself, simplifying the fasta record names to chr1, chrX, etc. This will be fixed in an upcoming release. Sorry for the inconvenience.