Tophat2.0.8b segment-junc err=1

[mattia]

Hi,
Last week I upgraded last version of Tophat; so now I'm using Tophat 2.0.8b with bowtie 1.0.0 (I need to align color space reads).
Before I used Tophat 1.4.1 and bowtie 0.12.8, and I never have any problems.

So I launch this command (the same of previous version except for --bowtie1) :

tophat -C --bowtie1 --no-coverage-search --library-type fr-secondstrand -G $annotation_file -g 10 -p 12 -r 90 -o $output_path $reference_file $csfastaF3 $csfastaF5

with reference_file="/.../ENSEMBL_HG19/Homo_sapiens.GRCh37.71.dna.23chr_c"

So, now this message appers:

[2013-05-06 16:04:37] Beginning TopHat run (v2.0.8b)
-----------------------------------------------
[2013-05-06 16:04:37] Checking for Bowtie
Bowtie version: 1.0.0.0
[2013-05-06 16:04:37] Checking for Samtools
Samtools version: 0.1.18.0
[2013-05-06 16:04:37] Checking for Bowtie index files
[2013-05-06 16:04:37] Checking for reference FASTA file
Warning: Could not find FASTA file /home/references/new_Human_annotations/ENSEMBL_HG19/Homo_sapiens.GRCh37.71.dna.23chr_c.fa
[2013-05-06 16:04:37] Reconstituting reference FASTA file from Bowtie index
Executing: ....

....
....

[2013-05-06 16:06:57] Reading known junctions from GTF file
[2013-05-06 16:07:11] Preparing reads

WARNING: read pairing issues detected (check prep_reads.log) !

left reads: min. length=75, max. length=75, 49812297 kept reads (187703 discarded)
right reads: min. length=35, max. length=35, 49679413 kept reads (320587 discarded)
[2013-05-06 16:20:43] Creating transcriptome data files..
[2013-05-06 16:21:59] Building Bowtie index from Homo_sapiens.GRCh37.71.23chr.fa
[2013-05-06 16:52:43] Mapping left_kept_reads to transcriptome Homo_sapiens.GRCh37.71.23chr with Bowtie
[2013-05-06 17:07:08] Mapping right_kept_reads to transcriptome Homo_sapiens.GRCh37.71.23chr with Bowtie
[2013-05-06 17:22:15] Resuming TopHat pipeline with unmapped reads
[2013-05-06 17:22:16] Mapping left_kept_reads.m2g_um to genome Homo_sapiens.GRCh37.71.dna.23chr_c with Bowtie
[2013-05-06 18:19:33] Mapping left_kept_reads.m2g_um_seg1 to genome Homo_sapiens.GRCh37.71.dna.23chr_c with Bowtie (1/3)
[2013-05-06 18:53:33] Mapping left_kept_reads.m2g_um_seg2 to genome Homo_sapiens.GRCh37.71.dna.23chr_c with Bowtie (2/3)
[2013-05-06 19:28:16] Mapping left_kept_reads.m2g_um_seg3 to genome Homo_sapiens.GRCh37.71.dna.23chr_c with Bowtie (3/3)
[2013-05-06 20:01:48] Mapping right_kept_reads.m2g_um to genome Homo_sapiens.GRCh37.71.dna.23chr_c with Bowtie
[2013-05-06 20:21:52] Mapping right_kept_reads.m2g_um_seg1 to genome Homo_sapiens.GRCh37.71.dna.23chr_c with Bowtie (1/1)
[2013-05-06 20:26:56] Searching for junctions via segment mapping
[FAILED]
Error: segment-based junction search failed with err =1
Error: could not get read# 11562180 from stream!

Any suggestions (maybe also from autors)?

Thank a lot.

[Jeremy]

Try and check read# 11562180, maybe something is wrong with the input files.

[mattia]

ok, I can check but I'm doubtful yet; I tested Tophat2, by using the same input file I've already aligned with Tophat1 (without any problem).

[mattia]

I check: All my reads are OK, but the problem still remains.

[mattia]

SOLVED!!! Tophat2.0.8b segment-junc err=1

I solve my problem: I suppose it can be a bug (in all tophat 2.0.x versions). You have not to set -p parameter. In this way Tophat works correctly, but, obviously, it takes too long!!!!
For paired-end alignment of 50 Mln of reads (Human) in color space:
Tophat 1.4.1 (-p 12) takes ~ 5 hours;
Tophat 2.0.8b (-p not set; default -p 1) takes ~20 hours.

[faltimiras]

I still having problems without the -p option...I think is because the library_type option..maybe??
I have paired-end reads (75b f and 35b rev) from SOLID 5500xl..

I got this message when I try to use tophat (v 1.4.1 and laters):

WARNING: read pairing issues detected (check prep_reads.log) !

Could u help me?

[GenoMax]

Have you looked at the "prep_reads.log" file to see if that has any additional information as the warning message indicates?

[faltimiras]

When I open the prep_reads.log file I have this:

prep_reads v1.4.1 (exported)
---------------------------
3906264 out of 3906264 reads have been filtered out

The reads are in colorspaced format (from SOLID 5500xl)...

The usage to run tophat 1.4.1 is as follow:

tophat -C --library-type fr-secondstrand -G file.gff3 -g 10 -r 12 -o output reference_index_dir f3.csfasta f5.csfasta f3.QV.qual f5.QV.qual

[andylemire]

When I open the prep_reads.log file I have this:

prep_reads v1.4.1 (exported)
---------------------------
3906264 out of 3906264 reads have been filtered out

The reads are in colorspaced format (from SOLID 5500xl)...

The usage to run tophat 1.4.1 is as follow:

tophat -C --library-type fr-secondstrand -G file.gff3 -g 10 -r 12 -o output reference_index_dir f3.csfasta f5.csfasta f3.QV.qual f5.QV.qual

Can you post the first few lines from each of the csfastas and qual files?

I think you have to specify -Q/--quals as well, so it knows to expect qual values in a separate file. Just guessing here (I stopped working in colorspace years ago), but I think that without that flag it treats the csfastas as fastqs, so it's possibly looking for readIDs to start with "@" instead of ">".

Something like:

tophat -C --quals --library-type fr-secondstrand -G file.gff3 -g 10 -r 12 -o output reference_index_dir f3.csfasta f5.csfasta f3.QV.qual f5.QV.qual

Also, do you really expect only 12bp inner mate distance with this library (-r 12)? Just curious...