Tophat v1.1.4 potential error with sam to bam conversion?

[jb2]

Using tophat v1.1.4, I have run into some issues with running tophat on the Illumina Human Body Map data. Here is some information from the log files from the run.

---
Last few lines of output from the tophat run:

[Sat Jan 8 02:13:21 2011] Mapping reads against segment_juncs with Bowtie
[Sat Jan 8 02:16:29 2011] Joining segment hits
[Sat Jan 8 02:22:00 2011] Reporting output tracks
Error: could not convert to BAM with samtools
---


---
Last line from run.log:

samtools view -S -b ./trim_to_75_trim_s_8_sequence_end1_tophat_out//tmp/accepted_hits.sam > ./trim_to_75_trim_s_8_sequence_end1_tophat_out//tmp/file9j1TKZ
---


---
Error message From accepted_hits_sam_to_bam.log:

[samopen] SAM header is present: 25 sequences.
Parse error at line 1129818: CIGAR and sequence length are inconsistent
---

Any idea what is going on and/or how I should go about solving it?

Edit:
Another thing I want to point out is that I saw that there was an accepted_hits.sam file in the tmp folder that remained after the tophat runs failed to complete. I tried to run this in Cufflinks instead and was getting errors that the sam file was not ordered correctly. I am posting this information in case that can help with understanding what might be happening with this issue.

[luxmare]

Hi jb2,

Sometimes, I also faced to the same problem. The latest version of TopHat often report erroneous read alignments. The problem in sam to bam file conversion seems to be caused by the erroneous CIGAR strings.

In my case, I mapped 76bp single-end reads to a reference sequence and the error in sam->bam conversion was occurred in the following line in tmp/accepted_hits.sam file. The CIGAR string means that this read has 536,870,957 bp in length.

XXXXX:3:15:3749:16676#0 0 chr1 21668580 3 28M47N536870907M43N22M *
0 0 GGCGTGTATTTGGGTTGAAGTTAAGCAACTGGTTCATGGACTGTG GGGGGGGGGGGGGGGGGGGGGGGDFFFEEAF=FDEEDE?DDAEBE NM:i:1
XS:A:- NH:i:2 CC:Z:= CP:i:21668580

So, in my analysis, I removed such lines with erroneous CIGAR strings in SMA file by checking the discrepancies in length between CIGAR and read. After that, I manually convert sam to bam and sort bam file by using SAMtools.

[Jon_Keats]

This isn't an issue with samtools version? I'll have to check the exact version but I know our recent version of Tophat did not work with the most recent samtools version as tophat would not recognize the version with a letter such as the current version samtools-0.1.12a.

[jb2]

Hi jb2,

Sometimes, I also faced to the same problem. The latest version of TopHat often report erroneous read alignments. The problem in sam to bam file conversion seems to be caused by the erroneous CIGAR strings.

In my case, I mapped 76bp single-end reads to a reference sequence and the error in sam->bam conversion was occurred in the following line in tmp/accepted_hits.sam file. The CIGAR string means that this read has 536,870,957 bp in length.




So, in my analysis, I removed such lines with erroneous CIGAR strings in SMA file by checking the discrepancies in length between CIGAR and read. After that, I manually convert sam to bam and sort bam file by using SAMtools.

Thanks for your help on this. Is there a quick tool for removing lines with problematic cigar strings? I'm sure I could throw a script together in perl pretty quickly, otherwise, but just curious.

[luxmare]

To remove erroneous lines from TopHat SAM file, I wrote a perl script by myself. But it's not quick tool. I also want to have a quick tool for that.

This isn't an issue with samtools version? I'll have to check the exact version but I know our recent version of Tophat did not work with the most recent samtools version as tophat would not recognize the version with a letter such as the current version samtools-0.1.12a.

Also in our system, the latest version of TopHat (v1.1.4) does not work with the latest version of SAMtools (v0.1.12a). We may have to use old version of SAMtools (v0.1.11).

[dnusol]

Hi, this is an old thread but I am still finding the same error with the newest versions.

I am using Tophat 1.3.1, Bowtie 0.12.7, Cufflinks 1.1.0 and samtools 0.1.16 and still need to convert the accepted_hits.bam to sam format, otherwise I get the

SAM error on line 9454: CIGAR op has zero length


Is there a compatibility issue?


Thanks

Dave

[pd]

CIGAR and Sequence length are inconsistent

Dear,

I am trying to convert sam file to bam file. I am pretty new to this conversion. But i got the following error :CIGAR and SEQUENCE length are inconsistent.
Is there a quick tool for removing lines with problematic cigar strings? because my sam file is almost 30 GB and doing it through perl script could be time consuming. And even if i remove these problemetic lines, Do i have to take care of some header lines (beginning of file) or i can remove them simply?