segment-based junction search failed

[Wei-HD]

Hi All,

I met a problem when I run the tophat, there was an error: segment-based junction search failed with err = 1. Therefore, there is no juction.bed file and wig file. Can someone give me some hints what mistakes I have made?

Thanks!

[204:~/desktop/wei/data] alan% ls
Bic.txt control.txt.zip mm9.2.ebwt mm9.fa tophat_out
Bic.txt.zip make_mm9.sh mm9.3.ebwt mm9.rev.1.ebwt
control.txt mm9.1.ebwt mm9.4.ebwt mm9.rev.2.ebwt
[204:~/desktop/wei/data] alan% tophat --solexa1.3-quals mm9 Bic.txt
Fri Jan 29 10:16:29 2010] Beginning TopHat run (v1.0.11)
-----------------------------------------------
[Fri Jan 29 10:16:29 2010] Preparing output location ./tophat_out/
[Fri Jan 29 10:16:29 2010] Checking for Bowtie index files
[Fri Jan 29 10:16:29 2010] Checking for reference FASTA file
[Fri Jan 29 10:16:29 2010] Checking for Bowtie
Bowtie version: 0.12.1.0
[Fri Jan 29 10:16:29 2010] Checking reads

seed length: 76bp
format: fastq
quality scale: --solexa1.3-quals
[Fri Jan 29 10:27:01 2010] Mapping reads against mm9 with Bowtie
[Fri Jan 29 12:28:40 2010] Joining segment hits
Splitting reads into 3 segments
[Fri Jan 29 12:39:19 2010] Mapping reads against mm9 with Bowtie
[Fri Jan 29 14:01:04 2010] Mapping reads against mm9 with Bowtie
[Fri Jan 29 15:20:38 2010] Mapping reads against mm9 with Bowtie
[Fri Jan 29 16:42:19 2010] Searching for junctions via segment mapping
[FAILED]
Error: segment-based junction search failed with err = 1
[204:~/desktop/wei/data] alan%

[Wei-HD]

Is that because Bowtie index's reference records have spaces in their names? I just checked the previous posts.

[Cole Trapnell]

It looks to me like there's some problem with the FASTQ file. Unless you ask TopHat to change the segment length to 76bp, it should default to 25bp here. TopHat has a bug currently that requires all of the sequence for each read to be on a single line (and same goes for the quality string). This will be fixed at some point, but I haven't had time recently to get to it.

[Wei-HD]

Thank you very much, Cole!

I might know the problem of my index file. Since there is no fasta (fa.) file in the downloaded mouse index file, but Tophat will use bowtie to build this file and save it to the output directory. And this step will take some time to finish, but I moved this file out of the Tophat_output file when Tophat is still running at this step. Therefore, the unfinished fasta file is 0 Kb.

[DanWebster]

Hello all,

I got the same error, but it occurred because I was also running another script (seqgene's snp function) that also drew on the hg18.fa file at the same time that I was running tophat.

Seems like it would be a rare thing, but just in case someone else had the same problem.

Followup: this also happened to me when I was running multiple tophat calls at once within the same linux instance. Though I can't know for sure that it was the same reason for failure, it seems like one tophat call "won" and got access to the file, and completed its run, while the other failed.

[trickytank]

Or in my case, my fasta file was a soft link, but the link was broken.

[paulorapazote]

Threads problem??

Hi

I had similar problem, but strangely (TopHat version 2.0.7) the error message only occurred when running the job in a node with 4 cores ( “-p 4” --num-threads option), and never with 16 cores (-p 16). So, In my case I believe in related with the number of threads used to run TopHat.

I hope could help.

Cheers,

P