bowtie2 malformed sam/bam files

Question

[Damian Kao]

Does anyone have any experience with bowtie2 (2.3.3) generating malformed sam/bam files? I am mapping a set of fasta files (not fastq) to a set of transcriptome sequences. I noticed from my .sam output that there were a lot of malformed entries.

I wrote a quick script to pull some of these lines out:

length of seq and qual are different:

D3NH4HQ1:59:C04PRACXX:3:1101:7947:37701/1       256     a3407467;22     25      40      25M75S  *       0       0       CGATCTGCCACAAAGGAAAGTAGAAGCGGTGAGTTCATCACCTAAAACTGGCAGCATTCAGGCTAATTTGCCAGAATCTTTCTCCGTAACAGGTGGCACT    IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII       AS:i:46 XS:i:46 XN:i:0  XM:i:0  XO:i:0  XG:i:0  NMMD:Z:25       YT:Z:UU

line-breaks inserted at wrong place. There were some lines with literally just these 3 characters which I assume is a line break inserted wrongly in a previous entry.

:UU

tab inserted in the wrong place. A tab is missing between the 9th and 10th column (the sequence starts with 0AAAC..):

D3NH4HQ1:59:C04PRACXX:3:1107:11276:12685/1      16      a11601010;43    579     40      36M1D64M        *       0       0AAACAGTAACAAGCACTCCTTGCTGGATTGAACTGCATCTGAATGGTCCTCTTCAATGGTTGGACAAAGTATTGACTCAGATGGGCTCTCCTTCAG       IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII    AS:i:192        XS:i:58 XN:i:0  XM:i:1  XO:i:0  XG:i:0  NM:i:1  MD:Z:56A43      YT:Z:UU

Is this due to my fasta file being wrong somehow? Maybe there is something wrong with bowtie2 threading; it is not pooling results from multiple threads and outputting them correctly?

Answer 1

[Damian Kao]

After dealing with this for several hours I realized the problem is with my bowtie2 index. I generated my index with bowtie2 2.3.0.

It looks like mapping with 2.3.3 on a 2.3.0 index is causing these problems.

[GenoMax]

Should the problem be considered to be with 2.3.3? I assume the old index works with v.2.3.0 correctly?

Index issues may be the hardest to track down. Did the release notes indicate a need to re-create the index? If not then does this qualify as a bug? On a different note, I think STAR may be the only program that recommends index recreation (am not a regular STAR user but I seem to recollect that advice).

[Damian Kao]

I am actually not sure if 2.3.0 works with 2.3.0 index. I would assume so?

The release notes doesn't indicate that a new index should be generated. I'll inform them of the bug.