solid reads stuck at searching juntions in TopHat

[zorph]

So I have installed Tophat v1.4 and I have run several illumina runs on it no problem using all 3 library parameters.

However, I have tried running SOLiD reads on it and I keep geting stuck at this section

[Tue Nov 30 09:05:31 2010] Beginning TopHat run (v1.1.4)
-----------------------------------------------
[Tue Nov 30 09:05:31 2010] Preparing output location /r/solid_xfer/Internal/AD/M_KOHLI/AD2/TopHat_8//
[Tue Nov 30 09:05:31 2010] Checking for Bowtie index files
[Tue Nov 30 09:05:31 2010] Checking for reference FASTA file
[Tue Nov 30 09:05:31 2010] Checking for Bowtie
Bowtie version: 0.12.7.0
[Tue Nov 30 09:05:31 2010] Checking for Samtools
Samtools version: 0.1.8.0
[Tue Nov 30 09:07:26 2010] Checking reads
min read length: 50bp, max read length: 50bp
format: fasta
[Tue Nov 30 10:30:43 2010] Mapping reads against hg18_c with Bowtie
[Tue Nov 30 13:25:41 2010] Joining segment hits
[Tue Nov 30 15:08:31 2010] Mapping reads against hg18_c with Bowtie(1/2)
[Tue Nov 30 17:46:39 2010] Mapping reads against hg18_c with Bowtie(2/2)
[Tue Nov 30 20:14:52 2010] Searching for junctions via segment mapping

does anyone have any suggestions? Thank you in advance.

[KevinLam]

I have the same problem as well with solid reads but I did get to finish it with a smaller number of reads. (no experience with illumina reads though). So I just assumed that junction mapping takes very long.
Is it possible for me to parallelize the junction mapping?

[mastal]

solid reads stuck at searching juntions in TopHat

Hi,

I am also having similar problems with TopHat. I have 50 bp SE SOLiD RNA-Seq data.
TopHat exited from the segment mapping step with error = -9.

Does anyone know what this error code means? Is it a memory problem?

I was running TopHat with the all the default parameters except for -p4, on a Linux machine (Ubuntu Maverick) with 7.5 Gb of memory. The .csfasta file I am trying to analyse has 92 million reads.

TopHat output:


[Sun Dec 5 17:03:44 2010] Beginning TopHat run (v1.1.4)
-----------------------------------------------
[Sun Dec 5 17:03:44 2010] Preparing output location ./tophat_out/
[Sun Dec 5 17:03:44 2010] Checking for Bowtie index files
[Sun Dec 5 17:03:44 2010] Checking for reference FASTA file
Warning: Could not find FASTA file /mnt/bowtie_color_genome/mm9_c.fa
[Sun Dec 5 17:03:44 2010] Reconstituting reference FASTA file from Bowtie index
[Sun Dec 5 17:06:57 2010] Checking for Bowtie
Bowtie version: 0.12.7.0
[Sun Dec 5 17:06:57 2010] Checking for Samtools
Samtools version: 0.1.12.0
[Sun Dec 5 17:07:30 2010] Checking reads
min read length: 50bp, max read length: 50bp
format: fasta
[Sun Dec 5 17:47:02 2010] Mapping reads against mm9_c with Bowtie
[Mon Dec 6 00:02:51 2010] Joining segment hits
[Mon Dec 6 01:07:59 2010] Mapping reads against mm9_c with Bowtie(1/2)
[Mon Dec 6 06:04:29 2010] Mapping reads against mm9_c with Bowtie(2/2)
[Mon Dec 6 10:40:53 2010] Searching for junctions via segment mapping
[FAILED]
Error: segment-based junction search failed with err = -9



Thanks,
Maria

[Daehwan]

I'm sorry for the problem, would any of you mind providing me with data set that cause the problem? (infphilo at cs.umd.edu) or you can reduce the number of reads while having the problem and give me the smaller data set.

Thanks,
Daehwan

[zorph]

Daehwan,
Has anyone sent you their file? if not, i can.

I would have to reduce the number of reads to send to you as my file is 31G/data set. What size should I send you and which files would you like?

[Daehwan]

@zorph,

I need a data set that gives the same problem, is it possible for you to open ftp session for your data, I'll just use the data for only test-purpose and delete once I'm done with fixing the problem. Please send me an email (infphilo at cs.umd.edu) and give me some description about your system and your TopHat command.

[holywoool]

Hi,
It seems that I've met another tough problem:
My tophat doesn't mapping reads to ref_seq,and doesn't give a .bed file. Astonishedly,the .sam file contains only headers:

@HD VN:1.0 SO:sorted
@SQ SN:chr1 LN:229974691
@SQ SN:chr10 LN:135001995
@SQ SN:chr11 LN:134204764
@SQ SN:chr12 LN:135371336
@SQ SN:chr13 LN:115868456
@SQ SN:chr14 LN:107349158
@SQ SN:chr15 LN:100063422
@SQ SN:chr16 LN:90682376
@SQ SN:chr17 LN:83384210
@SQ SN:chr18 LN:77261746
@SQ SN:chr19 LN:64473437
@SQ SN:chr20 LN:62293572
@SQ SN:chr21 LN:46489110
@SQ SN:chr22 LN:50165558
@SQ SN:chr2a LN:114460064
@SQ SN:chr2b LN:248603653
@SQ SN:chr3 LN:203962478
@SQ SN:chr4 LN:194897272
@SQ SN:chr5 LN:183994906
@SQ SN:chr6 LN:173908612
@SQ SN:chr7 LN:160261443
@SQ SN:chr8 LN:145085868
@SQ SN:chr9 LN:138509991
@SQ SN:chrX LN:155361357
@SQ SN:chrY LN:23952694
@PG ID:TopHat VN:1.1.4 CL:/home/wdq/tophat-1.1.4.Linux_x86_64/tophat --color --quals -p 2 -G OUT.txt genome_o_6 part.csfasta part_QV.qual

[root@localhost tophat-1.1.4.Linux_x86_64]# tophat --color --quals -p 2 -G OUT.txt genome_o_6 part.csfasta part_QV.qual

[Wed Dec 8 15:28:35 2010] Beginning TopHat run (v1.1.4)
-----------------------------------------------
[Wed Dec 8 15:28:35 2010] Preparing output location ./tophat_out/
[Wed Dec 8 15:28:35 2010] Checking for Bowtie index files
[Wed Dec 8 15:28:35 2010] Checking for reference FASTA file
[Wed Dec 8 15:28:35 2010] Checking for Bowtie
Bowtie version: 0.12.7.0
[Wed Dec 8 15:28:35 2010] Checking for Samtools
Samtools version: 0.1.11.0
[Wed Dec 8 15:29:13 2010] Checking reads
min read length: 50bp, max read length: 50bp
format: fasta
[Wed Dec 8 15:29:16 2010] Reading known junctions from GTF file

What made it come to this?Is anyone have the same problem?

[mastal]

solid reads stuck at searching juntions in TopHat

My problem seems to have been solved by running tophat on a machine with more memory.

When I got the error 'segment-based junction search failed with err = -9' I was using a machine with 7.5 Gb of memory.

I created smaller files (csfasta and qual) with only 5 million reads, and that ran without any problem.

I then ran tophat on a 15 Gb machine on the files with 92 million reads, and it completed the run without any problems in about 36 hours, and it was using all of the memory.

@holywoooi
From your output, it seems that tophat has stopped at the stage where it is reading the junctions file, so it isn't surprising that there is only the header in the sam file.


Maria

[KevinLam]

My problem seems to have been solved by running tophat on a machine with more memory.

When I got the error 'segment-based junction search failed with err = -9' I was using a machine with 7.5 Gb of memory.

I created smaller files (csfasta and qual) with only 5 million reads, and that ran without any problem.

I then ran tophat on a 15 Gb machine on the files with 92 million reads, and it completed the run without any problems in about 36 hours, and it was using all of the memory.

@holywoooi
From your output, it seems that tophat has stopped at the stage where it is reading the junctions file, so it isn't surprising that there is only the header in the sam file.


Maria

I was hoping that Tophat wouldn't require much more than Desktop memory size.

I am new to the pipeline would you know if it is ok to combine the reads after splitting them into 5 mil reads?