Tophat/Cufflinks using own GFF3 file problems

[James]

Hi people.

I'm having a couple of problems with aligning using my own junctions in a GFF3 file. I'm just starting out with this just so I can make this simple and just look at changes in gene expression before I get into new transcripts and things like that. So I am using a GFF3 file that contains the templates. Also when I attempt to use --no-novel-juncs option so tophat won't look for novel splicing. I get this error:

tophat -G GFF3/data/gff3/combined.gff --no-novel-juncs indexes/genomic reads/11.3.10/R43s_4_sequence.fastq

[Sun Jul 4 16:24:40 2010] Beginning TopHat run (v1.0.11)
-----------------------------------------------
[Sun Jul 4 16:24:40 2010] Preparing output location ./tophat_out/
[Sun Jul 4 16:24:40 2010] Checking for Bowtie index files
[Sun Jul 4 16:24:40 2010] Checking for reference FASTA file
[Sun Jul 4 16:24:40 2010] Checking for Bowtie
Bowtie version: 0.12.3.0
[Sun Jul 4 16:24:40 2010] Checking reads
seed length: 36bp
format: fastq
quality scale: --phred33-quals
[Sun Jul 4 16:26:29 2010] Reading known junctions from GFF file
[Sun Jul 4 16:27:31 2010] Mapping reads against DictyAx4_genomic with Bowtie
[Sun Jul 4 16:46:16 2010] Joining segment hits
[Sun Jul 4 16:48:28 2010] Retrieving sequences for splices
[Sun Jul 4 16:48:32 2010] Indexing splices
Warning: Empty input file
Error: No unambiguous stretches of characters in the input. Aborting...
Command: bowtie-build ./tophat_out/tmp/segment_juncs.fa ./tophat_out/tmp/segment_juncs
[FAILED]
Error: Splice sequence indexing failed with err = 1

This works when i don't include the --no-novel-juncs option.

Is this a problem with the GFF3 file I'm using? should it be formatted differently?

Many thanks, James

[Bio.X2Y]

There's a space character in your file location on your command line ("genomic reads") - perhaps that's why tophat is reporting "Empty input file". Try renaming that directory and trying again.

[Bio.X2Y]

Actually, sorry, didn't notice you were referring to an index there ; ignore my comments!

[Bio.X2Y]

I'm not sure where you got your GFF3, but some others have had issues taking UCSC GTF files and converting them to GFF3 files that TopHat can use. As you probably know, TopHat can also have a list of raw junctions in another simpler format using the "-j/--raw-juncs" flag. I've attached a perl script on another post that creates one of these raw junction files from a UCSC annotation (http://seqanswers.com/forums/showthread.php?t=5254).

[James]

Hi,

I downloaded the GFF from the database for my organism

http://dictybase.org/Downloads/

take a look at the formatting?

Cheers, J

[Bio.X2Y]

I can see nothing wrong with the format after a quick look; perhaps you could try e-mailing Cole, he might be able to shed light on whether the format is the issue.