Reduce Sam/Bam File Size

Question

[Nicolas Rosewick]

Hi,

I've a little question on sam and bam file sizes.

When I use bwa on paired-end reads (~50M reads) on a small reference sequence (~100 kb) , I've a bam file of about 5 Go . After looking the alignment, only a few reads aligned on this reference (~500 reads max)

But When I use tophat with the same input and the same reference, the output bam has a size of only 10 kb and the number of aligned reads is the same...

So is it a way to reduce my bam file ?

Thanks

Answer 1

[Sean Davis]

Just guessing, but your first BAM file probably contains both aligned and unaligned reads. The Tophat-produced BAM file contains only aligned reads. Both are "correct" BAM files, but which is most useful will depend on your particular needs. If you decide that you do not need to have the unaligned reads, you can use samtools view with a flag filter to remove reads that are unmapped.

[Nicolas Rosewick]

Is it ok like this : samtools view -F 4 in.bam > out.bam

[Sean Davis]

You'll probably also want to include -b and -h for bam output and SAM header, respectively.