Bowtie mapped reads all unique

[tulia]

Hi,

I am aligning several different sets ChIP-Seq data from an illumina hiseq2500 with bowtie and have found that all the mapped reads are unique when I run SAMTOOLS. This seems a little suspicious.

This is what I have done so far:
mapped with bowtie:
./bowtie-0.12.7/bowtie -S --fr -p 6 --un unaligned.fastq \
mm9 \
sample.fastq \
accepted_hits.sam 2>filter.bowtie.err

extract the mapped reads with samtools:
/share/apps/samtools/samtools view -Sb -F 4 accepted_hits.bam > mapped.bam

extract the unique reads:
samtools view -b -q 255 mapped.bam >unique.bam
also tried
samtools/samtools view -b -q 1 mapped.bam >unique.bam

The files where exactly the same for for the mapped.bam and the unique.bam either why I tried it. I then tried to use the bowtie flag -m 10 to include only unique reads and found that this was indeed smaller than my original mapping without the -m 10 flag. Mapping twice however wouldn't be a practical solution as I need to take extract the unique reads out later on.

I would appreciate any feedback on what may be causing all the mapped reads to be considered unique when I run samtools.

[TiborNagy]

Bowtie does not calculate mapping quality, so it is not a good measure:

http://seqanswers.com/forums/showthread.php?t=3142

[tulia]

Thanks for your reply.

[jgreenbaum]

I believe tulia meant to say '-m 1' instead of '-m 10'.

In any case, the link that you sent seems to reaffirm that that mapping qualities reported by bowtie are an indication of uniqueness, as does the bowtie manual. In the original mapping results (without specifying '-m 1'), the mapq of all reads is 255, and indeed all read IDs are unique. However, specifying '-m 1' produces a SAM file with only a subset of the output.

Can anyone explain what the '-m 1' parameter is doing differently than filtering for reads with a 255 mapq? Are we not understanding the parameter correctly?

Thanks,

J

[TiborNagy]

You are right, running bowtie with -m 1 gives unique alignments.