tophat2 errors

[ahmetz]

now that we have tophat2 out the door, i'm sure we'll have bunch of tophat2 related error messages and solutions - hopefully.

thank you to the team for releasing the new version. i'm very excited for the combination of tophat and tophat-fusion.

now my problem. I've downloaded and installed bowtie2 and tophat2. everything started out just fine until tophat tried to search for junctions via segment mapping. I got an error 5

tophat2 -o A673_rep1_tophat2 -r 158 --mate-std-dev 44 -p 24 --microexon-search --library-type fr-unstranded --b2-sensitive --fusion-search --fusion-min-dist 50000 --fusion-multireads 5 --fusion-multipairs 5 ~/genome_reference/bowtie_indexed/hg19.bowtie2 A673_rep1_R1.fastq A673_rep1_R2.fastq 

[2012-04-11 14:16:47] Beginning TopHat run (v2.0.0)
-----------------------------------------------
[2012-04-11 14:16:47] Checking for Bowtie
		  Bowtie version:	 2.0.0.5
[2012-04-11 14:16:47] Checking for Samtools
		Samtools version:	 0.1.18.0
[2012-04-11 14:16:47] Checking for Bowtie index files
[2012-04-11 14:16:47] Checking for reference FASTA file
	Warning: Could not find FASTA file /Users/zehira/genome_reference/bowtie_indexed/hg19.bowtie2.fa
[2012-04-11 14:16:47] Reconstituting reference FASTA file from Bowtie index
  Executing: /Users/zehira/local/bin/bowtie2-inspect /Users/zehira/genome_reference/bowtie_indexed/hg19.bowtie2 > A673_rep1_tophat2/tmp/hg19.bowtie2.fa
[2012-04-11 14:19:57] Generating SAM header for /Users/zehira/genome_reference/bowtie_indexed/hg19.bowtie2
	format:		 fastq
	quality scale:	 phred33 (default)
[2012-04-11 14:20:00] Preparing reads
	 left reads: min. length=75, count=12942699
	right reads: min. length=75, count=12927964
[2012-04-11 14:28:54] Mapping left_kept_reads against hg19.bowtie2 with Bowtie2 
[2012-04-11 14:44:43] Mapping left_kept_reads_seg1 against hg19.bowtie2 with Bowtie2 (1/3)
[2012-04-11 14:47:28] Mapping left_kept_reads_seg2 against hg19.bowtie2 with Bowtie2 (2/3)
[2012-04-11 14:50:06] Mapping left_kept_reads_seg3 against hg19.bowtie2 with Bowtie2 (3/3)
[2012-04-11 14:53:27] Mapping right_kept_reads against hg19.bowtie2 with Bowtie2 
[2012-04-11 15:09:38] Mapping right_kept_reads_seg1 against hg19.bowtie2 with Bowtie2 (1/3)
[2012-04-11 15:12:11] Mapping right_kept_reads_seg2 against hg19.bowtie2 with Bowtie2 (2/3)
[2012-04-11 15:14:44] Mapping right_kept_reads_seg3 against hg19.bowtie2 with Bowtie2 (3/3)
[2012-04-11 15:17:26] Searching for junctions via segment mapping
	[FAILED]
Error: segment-based junction search failed with err =-5
  Reason: image not found

I'm hoping it's an easy solution and am sure it's something i did wrong. anyone has any ideas?

Thanks!

[NateP]

I've had an error pop about at the same point when I gave it a test run the other day myself.
Was single end mapping.
Went through the left_kept_reads then:
[timestamp] Searching for junctions via segment mapping
[timestamp] Retrieving sequences for splices
[FAILED]
Error: Splice sequence retrieval failed with err =-6

[Jon_Keats]

Have you tried the fusion version with the -bowtie1 option as suggested when you are doing the fusion detection.

[ahmetz]

yeah, and i also removed fusion detection completely but they all end up giving the same error.

[Nicolas]

Same here. Tophat2 hangs forever at the same step "Searching for junctions via segment mapping", with paired-end HiSeq fastq reads (~140M reads), with or without --fusion-search option.
I ran it succesfully on a small subset of the same files (1M reads), so I suspect it has to deal with data size / memory...
Hope for a quick fix!

[plabaj]

I have a simillar problem at the same stage:
[FAILED]
Error: segment-based junction search failed with err =-6
Error: invalid 'B' array subtype (B)!

and TopHat was run with the following parameters:
--bowtie1 --report-secondary-alignments --bowtie-n

[gpertea]

There are quite a few different problems reported in this thread - I apologize for them. We are aware of a couple of these problems and we are trying to fix them, though we are encountering some difficulties reproducing some of these errors.
We just updated the packages on the website with a couple of last minute fixes so please re-download the tophat-2.0.0 package of your choice (yes, it's the same version, same file names) from http://tophat.cbcb.umd.edu and try running the new version again on your test data. If tophat 2 still fails at any stage in the process please report the failure at tophat.cufflinks@gmail.com - it would also help if you attach a tarball or zip file with the content of the "logs" directory when you post a bug report there.

Thank you,
-Geo

[NateP]

Originally Posted by gpertea

There are quite a few different problems reported in this thread - I apologize for them. We are aware of a couple of these problems and we are trying to fix them, though we are encountering some difficulties reproducing some of these errors.
We just updated the packages on the website with a couple of last minute fixes so please re-download the tophat-2.0.0 package of your choice (yes, it's the same version, same file names) from http://tophat.cbcb.umd.edu and try running the new version again on your test data. If tophat 2 still fails at any stage in the process please report the failure at tophat.cufflinks@gmail.com - it would also help if you attach a tarball or zip file with the content of the "logs" directory when you post a bug report there.

Thank you,
-Geo

The updated version fixed the error I posted earlier in the thread. Thank you for the quick work!

[westerman]

Yes, the updated version also solved the problems I was having with crashes and hangs in the 'Joining segment' part of the program. My samples are being mapped to a non-standard organism that has over 80 scaffolds.

[vyellapa]

I did'nt include the fusion flags and it worked ok. Is anybody seeing any significant speedup compared to 1.4.1? I dont.

time tophat2 -p 8 -r 50 --mate-std-dev 50 --transcriptome-index=./tophat2_transcript/GRCh37_E66 -o ./t2 \
-n 2 -N 4 --b2-sensitive --library-type fr-unstranded \
--rg-id ID000 --rg-sample M123 --rg-library lib0000 --rg-platform-unit IlluminaLn7 --rg-center TGen \
$RESOURCES/bowtie2/GRCh37.62 ./1mr1.fastq ./1mr2.fastq

[NicoBxl]

Tophat 2 also hangs forever in

[2012-04-13 03:41:33] Searching for junctions via segment mapping

and in the segment_juncs.log

>> Performing segment-search:
Loading left segment hits...

[DineshCyanam]

Yes... Tophat 2 is taking extremely long (~ 30 hrs)... particularly the Searching for junctions via segment mapping step as NicoBxl above observed. Here is my log.

$ tophat -r 150 --solexa1.3-quals -p 24 -g 20 --library-type fr-unstranded -o tophat_out $BOWTIE2_INDEXES/genome 61T4D/s_1_1_sequence.txt.gz,6286U/s_1_1_sequence.txt.gz 61T4D/s_1_2_sequence.txt.gz,6286U/s_1_2_sequence.txt.gz

[2012-04-13 18:14:51] Beginning TopHat run (v2.0.0)
-----------------------------------------------
[2012-04-13 18:14:51] Checking for Bowtie
Bowtie version: 2.0.0.5
[2012-04-13 18:14:51] Checking for Samtools
Samtools version: 0.1.18.0
[2012-04-13 18:14:51] Checking for Bowtie index files
[2012-04-13 18:14:51] Checking for reference FASTA file
[2012-04-13 18:14:51] Generating SAM header for /Homo_sapiens/Ensembl/GRCh37/Sequence/BowtieIndex/genome
format: fastq
quality scale: phred64 (reads generated with GA pipeline version >= 1.3)
[2012-04-13 18:14:53] Preparing reads
left reads: min. length=60, count=44338793
right reads: min. length=60, count=44098248
[2012-04-13 18:29:52] Mapping left_kept_reads against genome with Bowtie2
[2012-04-13 19:17:01] Mapping left_kept_reads_seg1 against genome with Bowtie2 (1/2)
[2012-04-13 19:28:48] Mapping left_kept_reads_seg2 against genome with Bowtie2 (2/2)
[2012-04-13 19:47:50] Mapping right_kept_reads against genome with Bowtie2
[2012-04-13 20:33:37] Mapping right_kept_reads_seg1 against genome with Bowtie2 (1/2)
[2012-04-13 20:47:07] Mapping right_kept_reads_seg2 against genome with Bowtie2 (2/2)
[2012-04-13 21:08:17] Searching for junctions via segment mapping
[2012-04-14 22:36:08] Retrieving sequences for splices
[2012-04-14 22:38:05] Indexing splices
[2012-04-14 22:51:34] Mapping left_kept_reads_seg1 against segment_juncs with Bowtie2 (1/2)
[2012-04-14 23:01:42] Mapping left_kept_reads_seg2 against segment_juncs with Bowtie2 (2/2)
[2012-04-14 23:15:09] Joining segment hits
[2012-04-14 23:37:53] Mapping right_kept_reads_seg1 against segment_juncs with Bowtie2 (1/2)
[2012-04-14 23:49:11] Mapping right_kept_reads_seg2 against segment_juncs with Bowtie2 (2/2)
[2012-04-15 00:04:38] Joining segment hits
[2012-04-15 00:28:13] Reporting output tracks
-----------------------------------------------
[2012-04-15 00:43:47] Run complete: 1 days 06:28:56 elapsed

Previous versions of Tophat for the same data took ~8 hours to run

[turnersd]

I'm also getting an error when using the --transcript-index option, "GFF transcripts file not found or invalid". The file exists, and it appears valid.

[DineshCyanam]

So I was able to reduce my runtime to ~4 hrs from ~30 hrs by using the --no-coverage-search and --no-novel-indels flags.

Originally Posted by DineshCyanam

Yes... Tophat 2 is taking extremely long (~ 30 hrs)... particularly the Searching for junctions via segment mapping step as NicoBxl above observed.
Previous versions of Tophat for the same data took ~8 hours to run

[Dario1984]

The location of the indexes doesn't work like the documentation says.

darstr@clark-lab:~/bowtie2$ echo $BOWTIE_INDEXES
/home/darstr/bowtie2/indexes/
darstr@clark-lab:~/bowtie2$ ls $BOWTIE_INDEXES
hg18.1.bt2  hg18.2.bt2  hg18.3.bt2  hg18.4.bt2  hg18.rev.1.bt2  hg18.rev.2.bt2

[2012-04-18 15:43:14] Beginning TopHat run (v2.0.0)
-----------------------------------------------
[2012-04-18 15:43:14] Checking for Bowtie
                  Bowtie version:        2.0.0.5
[2012-04-18 15:43:14] Checking for Samtools
                Samtools version:        0.1.18.0
[2012-04-18 15:43:14] Checking for Bowtie index files
Error: Could not find Bowtie 2 index files (hg18.*.bt2)

Not only is the environment variable ignored, but it also doesn't look in the indexes directory in the directory of the bowtie instance.

The only was it works is if I change directories to /home/darstr/bowtie2/indexes/ and run it from there.

[vyellapa]

Have you tried indexing your reference using

bowtie2-build <reference.fa> <bt2_base>

I ran into the same error when I downloaded the pre-built indexes, however, building my own worked ok.

[delagoya]

Originally Posted by gpertea

There are quite a few different problems reported in this thread - I apologize for them. We are aware of a couple of these problems and we are trying to fix them, though we are encountering some difficulties reproducing some of these errors.
We just updated the packages on the website with a couple of last minute fixes so please re-download the tophat-2.0.0 package of your choice (yes, it's the same version, same file names) from http://tophat.cbcb.umd.edu and try running the new version again on your test data. If tophat 2 still fails at any stage in the process please report the failure at tophat.cufflinks@gmail.com - it would also help if you attach a tarball or zip file with the content of the "logs" directory when you post a bug report there.

Thank you,
-Geo

I'm sorry, but what is the point of version numbers if you are not going to differentiate a bug fix release? Please reconsider and release it as 2.0.1. Even 2.0.0.1 would be OK.

Mistakes happen in all software. Rails is famous for releasing a version, and releasing another version 1 day later due to a major bug or security hole they found after the release.

For more on the topic, see http://semver.org/

[Dario1984]

I have a new problem. For one samples, the mapped reads between TopHat 1.33 and 2 went from 21 million to 12 million reads. I decided to BLAT some of the reads in the unmapped.bam file.

I wonder why reads like this are excluded ?

ACTIONS	QUERY	SCORE	START	END	QSIZE	IDENTITY	CHRO	STRAND	START	END	SPAN
browser details	YourSeq	76	1	76	76	100.0%	M	+	4951	5026	76
browser details	YourSeq	72	1	76	76	97.4%	1	+	555363	555438	76

The first alignment is better.

[dvanic]

Originally Posted by delagoya

I'm sorry, but what is the point of version numbers if you are not going to differentiate a bug fix release? Please reconsider and release it as 2.0.1. Even 2.0.0.1 would be OK.

Mistakes happen in all software. Rails is famous for releasing a version, and releasing another version 1 day later due to a major bug or security hole they found after the release.

For more on the topic, see http://semver.org/

+1 to this.
It's just that there are probably people who are using tophat who haven't checked this thread in the last week, which may mean they'll be using a version without this bug fixed.
Thank you in advance!

[labrat73]

Originally Posted by delagoya

I'm sorry, but what is the point of version numbers if you are not going to differentiate a bug fix release? Please reconsider and release it as 2.0.1. Even 2.0.0.1 would be OK.

Mistakes happen in all software. Rails is famous for releasing a version, and releasing another version 1 day later due to a major bug or security hole they found after the release.

For more on the topic, see http://semver.org/

+2. Also, I'm still hoping for a fix to the problem of segment_juncs process hanging for >24 hours. Saw a post earlier where user was able to shorten the process by not using certain flags. For now, back to using Tophat 1.4.0 and TopHatFusion.