Hi - Pls see below On 8/27/13 6:36 AM, Delong, Zhou wrote:

Hello, I have run several analysis with Tophat 2 on my local instance of galaxy and I get this error for all of them..

segment-based junction search failed with err = 1 or -9

Here is an example of full error report:

Error in tophat:

[2013-08-23 11:56:58] Beginning TopHat run (v2.0.6) ----------------------------------------------- [2013-08-23 11:56:58] Checking for Bowtie Bowtie version: 2.0.2.0 [2013-08-23 11:56:58] Checking for Samtools Samtools version: 0.1.18.0 [2013-08-23 11:56:58] Checking for Bowtie index files [2013-08-23 11:56:58] Checking for reference FASTA file [2013-08-23 11:56:58] Generating SAM header for /usr/local/data/bowtie2/hg19/hg19 format: fastq quality scale: phred33 (default) [2013-08-23 11:58:04] Preparing reads left reads: min. length=50, max. length=50, 145339247 kept reads (34946 discarded) right reads: min. length=50, max. length=50, 145340153 kept reads (34040 discarded) [2013-08-23 14:16:21] Mapping left_kept_reads to genome hg19 with Bowtie2 [2013-08-24 01:04:37] Mapping left_kept_reads_seg1 to genome hg19 with Bowtie2 (1/2) [2013-08-24 03:38:22] Mapping left_kept_reads_seg2 to genome hg19 with Bowtie2 (2/2) [2013-08-24 05:29:58] Mapping right_kept_reads to genome hg19 with Bowtie2 [2013-08-24 19:50:22] Mapping right_kept_reads_seg1 to genome hg19 with Bowtie2 (1/2) [2013-08-24 22:36:38] Mapping right_kept_reads_seg2 to genome hg19 with Bowtie2 (2/2) [2013-08-25 01:40:37] Searching for junctions via segment mapping Coverage-search algorithm is turned on, making this step very slow Please try running TopHat again with the option (--no-coverage-search) if this step takes too much time or memory. [FAILED] Error: segment-based junction search failed with err =-9 Collecting potential splice sites in islands

cp: cannot stat `/home/galaxy/galaxy-dist/database/job_working_directory/000/515/tophat_out/deletions.bed': No such file or directory cp: cannot stat `/home/galaxy/galaxy-dist/database/job_working_directory/000/515/tophat_out/insertions.bed': No such file or directory

I did some research on the internet and it seems to be a memory problem to me, is there any solution other than rerun these jobs on a more powerful machine? Or modify the parameters, as suggested in the error report. You might also be able to run the job with smaller inputs by breaking them up into smaller datasets (paired), but how appropriate this is depends on what you plan to do with the mapped data after.

And why has Bowtie/Tophat discard different numbers of reads? What will be the impact? Does it means that if I don't have exact matches between the paired end input, it is still be possible to run the job? Not having both ends of all pairs map in every experiment is probably expected. And you can certainly run the mapping job without even the initial FASTQ inputs having exactly the same set of matched pairs (for example: maybe some were filtered out differently during QA steps).

The impact of having matched pairs will manifest in later steps in your analysis. And it depends on what you are doing how this needs to be manipulated upfront (or not!). Some tools, such as the Tuxedo RNA-seq tools, will filter down to matched pairs during later steps, such as Cufflinks, on their own. If doing variant calling, certain tools will also filter down to matched pairs or have tool form options to limit analysis to matched pairs - but others will expect that only matched pairs are input from the start. There are tools in the SAMTools tool group to filter/merge mapping results and tools in the Picard tool group to add/replace/merge labels or data results - all as needed. Each tool has help on the form itself, including links to the external source documentation. Hopefully this helps, Jen Galaxy team

Thanks, Delong

___________________________________________________________ The Galaxy User list should be used for the discussion of Galaxy analysis and other features on the public server at usegalaxy.org. Please keep all replies on the list by using "reply all" in your mail client. For discussion of local Galaxy instances and the Galaxy source code, please use the Galaxy Development list:

http://lists.bx.psu.edu/listinfo/galaxy-dev

To manage your subscriptions to this and other Galaxy lists, please use the interface at:

http://lists.bx.psu.edu/

To search Galaxy mailing lists use the unified search at:

http://galaxyproject.org/search/mailinglists/

-- Jennifer Hillman-Jackson http://galaxyproject.org