samtools rmdup error

[lukas1848]

Hi,

I am trying to run samtools rmdup on a bam file and I get the following error messages.

[bam_rmdup_core] inconsistent BAM file for pair 'HWI-ST1027:65: D03FGACXX:4:2107:2691:8134'. Continue anyway.
[bam_rmdup_core] inconsistent BAM file for pair 'HWI-ST1027:65: D03FGACXX:4:2208:17576:79909'. Continue anyway.
[bam_rmdup_core] inconsistent BAM file for pair 'HWI-ST1027:65: D03FGACXX:4:2301:18313:189015'. Continue anyway.
[bam_rmdup_core] inconsistent BAM file for pair 'HWI-ST1027:65: D03FGACXX:4:2307:15133:98183'. Continue anyway.
[bam_rmdup_core] inconsistent BAM file for pair 'HWI-ST1027:65: D03FGACXX:4:2301:4254:2861'. Continue anyway.

I googled and didn't find anything on that, so I hope that someone here might be able to help me.


edit: just some additional info. I mapped 100 bp PE Illumina reads with stampy&bwa.

[airtime]

Hi,

did you sort your bam file before?
Some further proceeding steps would be helpful (like full command execution).

[lukas1848]

I sorted my bam files

samtools view -uS ACAGTG_001.sam | samtools sort - ACAGTG

and removed the pcr dups like this

samtools rmdup ACAGTG_001.bam ACAGTG_001_rmdup.bam

If I use the -S option in rmdup to force treating PE reads as SE reads, the error does not occur (which seems not really surprising, though)

[airtime]

Hi,

ok. How big are your files and your RAM? If you convert sam to uncompressed bam is their size similar?

regards Air

[lukas1848]

I used a m1.large EC2 instance, so I had 7.5 gb ram. The sam files are ~2.2 gb each. The uncompressed bams are 1/4th of the sam files' size.
Is this somehow suspicious? I mean, should uncompressed bams be of roughly the same size as the sam template?

I am thinking about trying out Picard's duplicate removal, but I want to make sure first that my bam files are correct.

[lukas1848]

I contacted the authors of stampy and they said that they occasionally record errors of that kind but that it does not show any ill effects. According to them, samtools is complaining without good reason.

I think I'll just go with that, even if it still is odd though.