Tophat 2.0.0 is stuck

[EGrassi]

Hello, I have a tophat process that seems stuck:

[2012-10-03 14:47:19] Beginning TopHat run (v2.0.0)
-----------------------------------------------
[2012-10-03 14:47:19] Checking for Bowtie
                  Bowtie version:        2.0.0.6
[2012-10-03 14:47:19] Checking for Samtools
                Samtools version:        0.1.18.0
[2012-10-03 14:47:19] Checking for Bowtie index files
[2012-10-03 14:47:19] Checking for Bowtie index files
[2012-10-03 14:47:19] Checking for reference FASTA file
[2012-10-03 14:47:19] Generating SAM header for ../alignment/expanded_genome2
        format:          fastq
        quality scale:   phred33 (default)
[2012-10-03 14:47:56] Reading known junctions from GTF file
[2012-10-03 14:48:01] Preparing reads
         left reads: min. length=50, count=29715271
        right reads: min. length=50, count=29535618
[2012-10-03 15:02:14] Using pre-built transcriptome index..
[2012-10-03 15:02:18] Mapping left_kept_reads against transcriptome genes with Bowtie2
[2012-10-03 15:18:27] Mapping right_kept_reads against transcriptome genes with Bowtie2
[2012-10-03 15:34:00] Converting left_kept_reads.m2g to genomic coordinates (map2gtf)
[2012-10-03 15:40:53] Converting right_kept_reads.m2g to genomic coordinates (map2gtf)
[2012-10-03 15:47:50] Resuming TopHat pipeline with unmapped reads
[2012-10-03 15:48:14] Mapping left_kept_reads.m2g_um against expanded_genome2 with Bowtie2
[2012-10-03 15:57:12] Mapping left_kept_reads.m2g_um_seg1 against expanded_genome2 with Bowtie2 (1/2)
[2012-10-03 16:00:07] Mapping left_kept_reads.m2g_um_seg2 against expanded_genome2 with Bowtie2 (2/2)
[2012-10-03 16:03:12] Mapping right_kept_reads.m2g_um against expanded_genome2 with Bowtie2
[2012-10-03 16:12:05] Mapping right_kept_reads.m2g_um_seg1 against expanded_genome2 with Bowtie2 (1/2)
[2012-10-03 16:15:47] Mapping right_kept_reads.m2g_um_seg2 against expanded_genome2 with Bowtie2 (2/2)
[2012-10-03 16:19:25] Searching for junctions via segment mapping
[2012-10-03 21:18:12] Retrieving sequences for splices
[2012-10-03 21:20:47] Indexing splices
[2012-10-03 21:33:14] Mapping left_kept_reads.m2g_um_seg1 against segment_juncs with Bowtie2 (1/2)

...since this step never took so long I'm beginning to wonder if maybe it is really stuck there...maybe there is a problem in the multicore approach. I see the process listed as running and using CPU but strace does not report any syscall and the virtual memory used is fixed...I am checking /proc/PID data but I am not sure which is the correct file to
understand this kind of problems (maybe stack?).
Anyone experienced this kind of problems? I don't see it listed as a solved issue in successive version so I am not sure if it's time-wise to stop it and try with a newer version.

Thank you very much,
E.G.

[vplagnol]

+1 to this. I had similar issues (perhaps the same one) with the tophat 2.0.4 and multicore. I now (painfully) only use single core and it works fine, albeit very slowly.

Vincent

[EGrassi]

Ouch, sad. Thanks, I will try to launch it in sigle core mode and see if it works (assuming that it will end before my contract does that ).

[billstevens]

Do you need to upgrade to Tophat 2? I got good alignment with Tophat 1.4 and Bowtie 1, and I'm wondering if I should upgrade. If it can only do one core at a time, I definitely won't.

[EGrassi]

...no luck. Even the single core process got stuck at that point...I'm working on a public dataset (GSE27003). I've added the --no-novel-juncs option and it has finished. I really do not understand what's the problem.

[turnersd]

I've had the same issues and my solution was to "upgrade" to Tophat 1.4.1.

[hbt]

We've also had problems running Tophat 2 on multiple cores, usually it would fail at the point where multiple BAM files are merged into 1.
However we found that running on 4 cores worked so have switched to that for now...

[EGrassi]

Tophat 2 on a different dataset worked for me with different number of cores, ranging from 2 to 6. I'm wondering what could be the problem, even if in this peculiar case the problem is still there with a single core.

[alaincoletta]

We're using TopHat2 on Ubuntu 12.0.4. Parts of the process take up the max number of threads (16 threads: -p 16). https://insilicodb.org