Tophat 1.0.7 problems

[jwaage]

Hi all,

I'm running TOPHAT 1.0.7 with 75bp solexa reads in a pipeline that worked in the previous version - this is my command-line:

tophat --p 3 --solexa-quals --max-gene-family 1 /Index/mm9 /Reads/reads.fq

And output:

Tue May 19 13:30:36 2009] Beginning TopHat run (v1.0.7)
-----------------------------------------------
[Tue May 19 13:30:36 2009] Preparing output location ./tophat_out/
[Tue May 19 13:30:36 2009] Checking for Bowtie index files
[Tue May 19 13:30:36 2009] Checking for reference FASTA file
[Tue May 19 13:30:36 2009] Checking for Bowtie
Bowtie version: 0.9.9.3
[Tue May 19 13:30:36 2009] Checking reads
seed length: 75bp
format: fastq
quality scale: solexa
Splitting reads into 3 segments
[Tue May 19 13:30:36 2009] Mapping reads against mm9 with Bowtie
[Tue May 19 13:33:44 2009] Mapping reads against mm9 with Bowtie
[Tue May 19 13:33:47 2009] Mapping reads against mm9 with Bowtie
[Tue May 19 13:33:50 2009] Searching for junctions via coverage islands
[FAILED]
Error: segment-based junction search failed

The segment_juncs.log ends with this line:

Loading left segment hits...Error: reference sequence for ID 1621485109
, not found, aborting

Any suggestions are appreciated,

Thank you in advance,

JW, Uni. of Copenhagen

[Cole Trapnell]

Hi Johannes,

Can you email me your logs directory? And in the future, when you get these sorts of errors, please just email me directly, as I'll see it more quickly.

Thanks,

Cole

[Thomas Doktor]

Hi,

I got the same error when running a test set (33 bp solexa single end reads) against a Bowtie index of chr5 made using a fasta reference with a somewhat long and complicated fasta-name including spaces, ">gi|37550270|ref|NT_034772.5|Hs5_34934 Homo sapiens chromosome 5 genomic contig, reference assembly". When I rebuilt the index using the hg19 chr5 reference with just ">chr5" for the fasta name I was able to run Tophat 1.0.7 without errors.

Of course this might not be the problem in your case, but perhaps it can help others.
Thomas

[jwaage]

That could be the cause of my problem, as my reference names are pretty long as well. Thanks for the suggestion.

Regards,
Johannes

[Cole Trapnell]

I've just posted a note on the TopHat site alerting users to this problem, and describing the workaround. As suggested above, the problem occurs when the Bowtie index's reference records have spaces in their names. This will be fixed in the next release. Thanks for your help, and your patience.