TopHat crashes when running bowtie

[Batool]

Hello all,

TopHat crashes on my files after apparently aligning reads with bowtie for some time.

Here is the error message:

####
foobar$ tophat -p 8 -o TOPHAT /refgenome/hg19 Sample1.fastq

[2013-02-01 17:15:03] Beginning TopHat run (v2.0.6)
-----------------------------------------------
[2013-02-01 17:15:03] Checking for Bowtie
Bowtie version: 2.0.4.0
[2013-02-01 17:15:03] Checking for Samtools
Samtools version: 0.1.18.0
Warning: we do not recommend to have both Bowtie1 and Bowtie2 indexes in the same directory
the genome sequence (*.fa) may not be compatible with one of them
[2013-02-01 17:15:03] Checking for reference FASTA file
format: fastq
quality scale: phred33 (default)
[2013-02-01 17:15:03] Preparing reads
left reads: min. length=202, max. length=202, 415678243 kept reads (42823 discarded)
[2013-02-01 23:07:52] Mapping left_kept_reads to genome hg19 with Bowtie2
[FAILED]
Error running bowtie:

foobar$
####

As you can see, the error message is not terribly informative regarding the nature of the problem.

Mysteriously, I am able to get TopHat to run just fine on a smaller file consisting of the first 100,000 reads of that same fastq file.

So it IS able to run fine given my current setup, just not on the full size file which granted is quite large...might this be the issue?

If anyone has any insight into this issue, I would greatly appreciate your thoughts and expertise.

Best regards,

Batool

[dGho]

I am new at running Tophat and bowtie, so my answer is probably something you have already tried. But in the documentation for bowtie, it said that bowtie1 and bowtie2 require different indexes. Is it possible that this error is due to the fact that you indexed the ref sequence using both bowtie1 and bowtie2? Have you tried separating them into different directories?

I think I would start with that.

If that doesn't help, have you considered trying to run bowtie1 and tophat1 to see if they work for you? I was having a bunch of errors using bowtie2, so I tried an older version of bowtie (and therefore had to switch to an older version of tophat) and I stopped having errors.

Sorry, that's the best advice I have...good luck

[earonesty]

Mysteriously as well, I am also having this issue where smaller samples complete, and some fail. Our tophat2 boxes have 60GB ram, and we're doing the run with --no-coverage-search. It I figure it out, I'll post it. I'm wondering if it's just the bowtie1/2 indexes in the same dir. (It seems very odd to me that someone would write a program that worked that way... i mean, bt2 is on all the files!)

I'm going to try reindexing. I indexed hg19 with bowtie2.0.6....but am running bowtie2.1.0

[sdriscoll]

it's a super odd warning - about the genome fasta maybe possibly being incompatible with one of the two references (bt1 vs bt2). it's like at some point they had somebody that built a bowtie1 index and at a later date overwrote the genome fasta with an updated version and built their bowtie2 index messing up the bowtie1 index's association with the genome's fasta source so they decided to code a warning in about that.

i like the mysterious "foobar$" error. Anybody looked at the bowtie source to see where that's happening? Actually - that's a good question. Can bowtie run the alignment of those reads without crashing? at least then we'd know if it's a tophat issue vs a bowtie issue.

[earonesty]

it's a super odd warning - about the genome fasta maybe possibly being incompatible with one of the two references (bt1 vs bt2). it's like at some point they had somebody that built a bowtie1 index and at a later date overwrote the genome fasta with an updated version and built their bowtie2 index messing up the bowtie1 index's association with the genome's fasta source so they decided to code a warning in about that.

i like the mysterious "foobar$" error. Anybody looked at the bowtie source to see where that's happening? Actually - that's a good question. Can bowtie run the alignment of those reads without crashing? at least then we'd know if it's a tophat issue vs a bowtie issue.

reindexing didn't help. foobar$ wasn't part of the error... just the prompt. testing "manually" running it, instead of through our grid engine, thinking it might be some ulimit thing

[ronaldrcutler]

try running this without using the '-p' flag

sometimes multithreading does not work due to keeping bowtie2 multithreading alignments in order