TopHat error: disk full

[joseph]

Hi All
I am analyzing one PE lane with read files 's_3_1_sequence.txt’ and 's_3_2_sequence.txt’;
here are the first lines the read files:

s_3_1_sequence.txt
@GAII:3:1:2:321#0/1
GGGGCCTGGGACTCTNGGTCCCCTACTGNAGACA
+GAII:3:1:2:321#0/1
`[`aaX`_aV`aaaZDTKT\X__^XGZZDVV``a
@GAII:3:1:2:314#0/1
CCACCAGGCGCCCGTNGTGGCGCAGGAANGGGTG
+GAII:3:1:2:314#0/1
_``aa_\\_\_aa_PDZVYZ\ZZPZ\TVDHZT\Z
@GAII:3:1:2:508#0/1
GTTCAGCAGGAATGCNGAGATCGGAAGANGGGTT

s_3_2_sequence.txt
@GAII:3:1:2:321#0/2
TCCCNCCTGCCCNNNGCTTCNNNGTTTTNNNTCA
+GAII:3:1:2:321#0/2
BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
@GAII:3:1:2:314#0/2
CAGTNCCAGCGCNNNAGCGTNNNGACCTNNNACC
+GAII:3:1:2:314#0/2
`_JJDZ_aBBBBBBBBBBBBBBBBBBBBBBBBBB
@GAII:3:1:2:508#0/2
TCATNCCTGCTTANNCTATANNNTAAGAGNNTCT
M1-80330:reads jdhahbi$

the command-line I used:
tophat -r 200 /mydir/bowtie-0.9.9.3/indexes/h_sapiens_asm s_3_1_sequence.txt s_3_2_sequence.txt

the output with the error is below; I checked the disk space and there are more than 100 GB available:

[Thu May 21 16:53:57 2009] Beginning TopHat run (v1.0.7)
-----------------------------------------------
[Thu May 21 16:53:57 2009] Preparing output location ./tophat_out/
[Thu May 21 16:53:57 2009] Checking for Bowtie index files
[Thu May 21 16:53:57 2009] Checking for reference FASTA file
[Thu May 21 16:53:57 2009] Checking for Bowtie
Bowtie version: 0.9.9.3
[Thu May 21 16:53:58 2009] Checking reads
seed length: 34bp
format: fastq
quality scale: phred
Splitting reads into 1 segments
[Thu May 21 17:00:49 2009] Mapping reads against h_sapiens_asm with Bowtie
Splitting reads into 1 segments
[Thu May 21 18:03:09 2009] Mapping reads against h_sapiens_asm with Bowtie
[Thu May 21 18:51:52 2009] Searching for junctions via coverage islands
[Thu May 21 18:59:12 2009] Searching for junctions via mate-pair closures
[Fri May 22 05:40:00 2009] Retrieving sequences for splices
[Fri May 22 05:48:53 2009] Indexing splices
Index is corrupt: File size for ./tophat_out/tmp/segment_juncs.1.ebwt should have been 3799224901 but is actually -495742395.
Please check if there is a problem with the disk or if disk is full.
[FAILED]
Error: Splice sequence indexing failed

Any suggestions are appreciated,
Thanks,

joseph

[Cole Trapnell]

Hi Joseph,

This is most likely the same bug as a few other users have reported, where with short, paired reads, it's possible for the splice index to become unreasonably large and that may be tripping Bowtie's index integrity checks. I have fixed this in my source tree, and the new version should be released next week. I am just tying up a few loose ends with the latest build.

Sorry for the inconvenience. If you'd like to test out a snapshot of the code to see if it resolves the problem for you, please email me directly.

[biterbilen]

Hi Cole,

I had a disk quote problem too. TopHat produced: >1.5TB!

I use v1.1.4 with the default setting to map ~1M SE total RNA reads which are 20-37 nt long. The huge file is produced by long_spanning_reads after junction mapping step.

Any ideas?
Thanks,

Biter

[Daehwan]

This is a bug due to variable read length, which we fixed, the next version will include the fix.