TopHat(1.3.2) reports error!help!

[holywoool]

Mapping RNA-Seq reads to reference with TopHat(v-0.1.3.2) results in error in that

Mapping right_kept_reads against genome_o_6 with bowtie
gzip:stdout:Broken pipe
...
10Error:[Error 2]No such file or directory:'./tophat_out/temp/right_kept_reads_missing.fq'

what's happened to gzip with right_kept_reads?The left_kept_reads was processed properly.
Need your help,thanks!

[Artur Jaroszewicz]

holywoool, have you found the answer to your question yet? I am getting the same error when I try to run Tophat. Does anybody know what is going on? Here is my entire message:

[Wed Nov 16 16:17:54 2011] Beginning TopHat run (v1.3.3)
-----------------------------------------------
[Wed Nov 16 16:17:54 2011] Preparing output location /u/home/mcdb/arturj/B1/
[Wed Nov 16 16:17:54 2011] Checking for Bowtie index files
[Wed Nov 16 16:17:54 2011] Checking for reference FASTA file
[Wed Nov 16 16:17:54 2011] Checking for Bowtie
Bowtie version: 0.12.5.0
[Wed Nov 16 16:17:54 2011] Checking for Samtools
Samtools Version: 0.1.18
[Wed Nov 16 16:17:54 2011] Generating SAM header for hg19_c
[Wed Nov 16 16:17:54 2011] Preparing reads
format: fastq
quality scale: phred64 (reads generated with GA pipeline version >= 1.3)
[Wed Nov 16 16:17:54 2011] Reading known junctions from GTF file
Left reads: min. length=49, count=164672067
[Wed Nov 16 17:01:35 2011] Mapping left_kept_reads against hg19_c with Bowtie
[Wed Nov 16 17:01:35 2011] Processing bowtie hits

gzip: stdout: Broken pipe
Traceback (most recent call last):
File "/u/home/mcdb/arturj/tophat-1.3.3.Linux_x86_64/tophat", line 2604, in ?
sys.exit(main())
File "/u/home/mcdb/arturj/tophat-1.3.3.Linux_x86_64/tophat", line 2563, in main
user_supplied_deletions)
File "/u/home/mcdb/arturj/tophat-1.3.3.Linux_x86_64/tophat", line 2218, in spliced_alignment
segment_len)
File "/u/home/mcdb/arturj/tophat-1.3.3.Linux_x86_64/tophat", line 1820, in split_reads
zreads = ZReader(reads_filename, params.system_params, False)
File "/u/home/mcdb/arturj/tophat-1.3.3.Linux_x86_64/tophat", line 1190, in __init__
self.file=open(filename)
IOError: [Errno 2] No such file or directory: '/u/home/mcdb/arturj/B1/tmp/left_kept_reads_missing.fq'

And the bash script calling it..

PATH=$PATH:/u/home/mcdb/arturj/bowtie-0.12.5/:/u/home/mcdb/arturj/tophat-1.3.3.Linux_x86_64/:/u/home/mcdb/arturj/samtools-0.1.18/
export PATH
export BOWTIE_INDEXES=/u/home/mcdb/arturj/bowtie-0.12.5/indexes/

tophat --solexa1.3-quals -g 1 -G /u/home/mcdb/arturj/hg19_refflat.gtf -p 8 -o /u/home/mcdb/arturj/B1 hg19_c /u/home/mcdb/arturj/B1/B1.fastq


Please help!

[holywoool]

Indeed,there needs a "-Q" option in the TopHat's command.Mine was done after adding "-Q".You may have a try.

[Artur Jaroszewicz]

That definitely does not help me. I tried it, despite the manual saying it's for separate quality files. Has anyone else had this problem?
Artur

[Artur Jaroszewicz]

And thank you for your response, holywoool!

[Ramprasad]

I guess tat means that there are no unaligned reads..
Why dont you simulate some reads and cat them to your read files and then try??

[Thomas Doktor]

Where do you see the release of version 1.3.1 and 1.3.3? They are downloadable from the site, but nothing about them is mentioned on the TopHat site.

[cjp]

Setting -g means you won't see any reads that align more than once as it sets both the -m and -k flags in Bowtie to 1:

-k <int> report up to <int> good alignments per read (default: 1)
-m <int> suppress all alignments if > <int> exist (def: no limit)

Chris

[Artur Jaroszewicz]

Originally Posted by Thomas Doktor

Where do you see the release of version 1.3.1 and 1.3.3? They are downloadable from the site, but nothing about them is mentioned on the TopHat site.

I see it only on the downloads site... http://tophat.cbcb.umd.edu/downloads/
It might be just a bug in this version, but I did try running 4 different datasets at a different time, and it worked fine. I've kept all the same options. The only thing that has changed is the data. The ONLY thing I can think of is that maybe it's an OOM error, just cleverly disguised. However, I'm running my data on 8 processors with 8 Gb each. Should be enough for a 30 Gb fastq file, no? After this try, I'll try going back to 1.3.1 or earlier.

cjp, I've been doing the -g option on all my data, and it's never not worked. I'm trying it without the -g argument, however. I'll let you know in a little bit if it works.

[Artur Jaroszewicz]

Still nothing.. Tried changing the annotation file from refflat to ensembl, tried removing the '-g' option, -Q (even though I don't see how that would work at all), and it's impossible for there to not be any unaligned transcripts with 164672067 49-bp reads. From my experience, only about half of the reads align anyway, so the probability of nothing aligning would be 2^(-8236033). Maybe I'll try using an earlier version or something, unless someone has any other suggestions?

[cjp]

-Q is usually for SOLiD colour-space.

When debugging failed TopHat runs, you can also try to run the individual commands from the command line yourself.

In the logs/ sub-directory of your output directory there should be a file called run.log which shows the commands that TopHat runs. There are also other log files in there - look to see if you can find errors (especially in the newest one or two files). Otherwise try running each command by itself from the directory you started TopHat in and see if and where empty files are made. The tmp stuff needed to do this is usually kept if a run fails.

Chris

[Artur Jaroszewicz]

Originally Posted by cjp

-Q is usually for SOLiD colour-space.

When debugging failed TopHat runs, you can also try to run the individual commands from the command line yourself.

In the logs/ sub-directory of your output directory there should be a file called run.log which shows the commands that TopHat runs. There are also other log files in there - look to see if you can find errors (especially in the newest one or two files). Otherwise try running each command by itself from the directory you started TopHat in and see if and where empty files are made. The tmp stuff needed to do this is usually kept if a run fails.

Chris

Excellent suggestion. I will try it once I get out of my Tryptophan coma

[Artur Jaroszewicz]

And once our computing cluster is up. Boo, maintenance!

[Artur Jaroszewicz]

Wowee. So I evidently was using the wrong index. I thought hg19_c was complete (it's prebuilt from the Bowtie website). I was supposed to use hg19.

Originally Posted by cjp

-Q is usually for SOLiD colour-space.

When debugging failed TopHat runs, you can also try to run the individual commands from the command line yourself.

In the logs/ sub-directory of your output directory there should be a file called run.log which shows the commands that TopHat runs. There are also other log files in there - look to see if you can find errors (especially in the newest one or two files). Otherwise try running each command by itself from the directory you started TopHat in and see if and where empty files are made. The tmp stuff needed to do this is usually kept if a run fails.

Chris

However, Chris, I learned a lot about debugging trying to find this error. Thank you greatly for the suggestion!