Bowtie signal 9 (KILL) error

[GeneJockey]

I'm aligning a bunch of Illumina runs using Bowtie2 and I'm getting a frustrating error with only a few of the fastq files.

I run them with the following command:

bowtie2 -p 10 -x mm10_genome -1 SRR594396_1.fastq -2 SRR594396_2.fastq --local --very-sensitive-local --no-mixed --no-discordant | samtools view -uS - | samtools sort -m 10000000000 - SRR594393_genome

Here are the outputs I get for a few fastq file pairs:

SRR594399_genome
[samopen] SAM header is present: 22 sequences.
Parse error at line 111594048: missing colon in auxiliary data
[bam_sort_core] merging from 2 files...

SRR594400_genome
[samopen] SAM header is present: 22 sequences.
bowtie2-align died with signal 9 (KILL)
[sam_read1] reference 'SRR59' is recognized as '*'.
[main_samview] truncated file.
[bam_sort_core] merging from 2 files...

SRR594401_genome
[samopen] SAM header is present: 22 sequences.
Parse error at line 111266936: sequence and quality are inconsistent
[bam_sort_core] merging from 2 files...

SRR594396_genome
[samopen] SAM header is present: 22 sequences.
bowtie2-align died with signal 9 (KILL)
Parse error at line 111938538: missing colon in auxiliary data
[bam_sort_core] merging from 2 files...

SRR594415_genome
[samopen] SAM header is present: 22 sequences.
bowtie2-align died with signal 9 (KILL)
Parse error at line 112131890: missing colon in auxiliary data
[bam_sort_core] merging from 2 files...

SRR594417_genome
[samopen] SAM header is present: 22 sequences.
bowtie2-align died with signal 9 (KILL)
Parse error at line 112044487: sequence and quality are inconsistent
[bam_sort_core] merging from 2 files...

I can't find anything about this KILL error and googling the parse errors isn't helping much. I think the parsing has something to do with bowtie failing and then the sam->bam operation subsequently failing on the last line.

Can anyone help with this KILL business?

[anthony_h]

Also getting KILL

I am also getting this when I turn on the -a option.

Without -a my file of ~7M reads gets mapped against human genome hg19 successfully. With -a it bombs out even if I truncate to 200k reads. Maybe the process is using too much memory and the OS is killing it?

[davidblaney]

I was thinking a memory issue when I first read this. Have a look at the requirement on there website. Failing that drop them an email.

[ctseto]

What happens when you use --local --very-sensitive-local at the same time? I assume it picks one and keeps on going?

[Richard Finney]

How much memory does you system have?

System may kill processes if there is severe resource starvation; i.e. no memory available.

[anthony_h]

Yes, it was running out of memory.

Switched to using -k instead of -a and it completed.

[jylee]

Hi all,
I am encountering same issue SIGNAL 9 (KILL). I wonder how you solved this issue.

Anthony h,
Could you describe more detail how you set -k <int> options?