tophat - no junctions

[crh]

Hi,

I've seen similar posts re: problems w/getting junctions from tophat recently. Here is my problem:

The test data supplied w/ tophat runs fine.

I rebuilt my genome index (bowtie-build) and then ran bowtie-inspect to generate a .fa file which I then used to replace the original .fa file. There was a file size diff between the original.fa used to generate the indexes and that generated by bowtie-inspect which I can't explain.

I created $BOWTIE_INDEX to set the full path to the index files.

I then ran:
tophat -o no_options $BOWTIE_INDEX/chlre4 ~/phosphate/soap/1690-P/run1/reads/s3.fastqsanger.gz


Again, no junctions.

[Sun Jul 24 14:30:08 2011] Beginning TopHat run (v1.3.0)
-----------------------------------------------
[Sun Jul 24 14:30:08 2011] Preparing output location no_options/
[Sun Jul 24 14:30:08 2011] Checking for Bowtie index files
[Sun Jul 24 14:30:08 2011] Checking for reference FASTA file
[Sun Jul 24 14:30:08 2011] Checking for Bowtie

Bowtie version: 0.12.7.0
[Sun Jul 24 14:30:08 2011] Checking for Samtools

Samtools Version: 0.1.16
[Sun Jul 24 14:30:08 2011] Generating SAM header for /seu/cs/home/project/binf/seq/bowtie_ind
ex/chlre4
[Sun Jul 24 14:30:08 2011] Preparing reads

format: fastq
quality scale: phred33 (default)
Left reads: min. length=35, count=8749834
[Sun Jul 24 14:32:51 2011] Mapping left_kept_reads against chlre4 with Bowtie
[Sun Jul 24 14:57:15 2011] Processing bowtie hits
[Sun Jul 24 15:01:05 2011] Searching for junctions via segment mapping

Warning: junction database is empty!
[Sun Jul 24 15:01:09 2011] Reporting output tracks
-----------------------------------------------
Run complete [00:36:54 elapsed]


Here is the data obtained (junctions.bed is empty):
-rw-r--r-- 1 charlesh student 325M 2011-07-24 15:07 accepted_hits.bam
-rw-r--r-- 1 charlesh student 52 2011-07-24 15:05 deletions.bed
-rw-r--r-- 1 charlesh student 54 2011-07-24 15:05 insertions.bed
-rw-r--r-- 1 charlesh student 52 2011-07-24 15:05 junctions.bed
-rw-r--r-- 1 charlesh student 0 2011-07-24 15:16 left_kept_reads.info


Any suggestions ?

thanks
Charles

[plassaaw]

Try using it with a .gtf file and the -G option within tophat. GTF files are available from ucsc.

[crh]

no junctions: problem resolved

Hi All,

The read set we are processing at 35nt in length.Based on comments from a colleague and the following SEQanswers thread (http://seqanswers.com/forums/showthread.php?t=8107), the segment-length parameter should be at most half the length of the total read length (e.g. 17).

I changed --segment-length to 17 and obtained junction predictions.

What is the minimum length sequence bowtie will map?

Another tophat question.
I ran with F=0 (default is -F=.15), to capture the low-expressing genes with the plan to post-filter the 'noise' from the resulting bed file.

Has anyone else done this, suggestions for post-filtering ?

thanks!

Charles