Samtools sort by name - bam size issue

Question

[quentin54520]

Hello all,

I want to sort my bam by queryname so i used the command:

Samtools sort -n -m 1 -@ 10 -o /Path/ouput.bam /path/input.bam

It's work fine but at the end the ouput.bam is really bigger thant input.bam (70Go vs 47Go). It is normal ? Before to go on with this bam i prefer be sure... My input bam is the ouput of samtools view (to remove unwanted read).

Thanks in advance,

Quentin

[ATpoint]

I would suggest to always be explicit when setting memory limits, so 1G rather than 1. Maybe they made it bullet-proof in recent samtools versions but (if memory serves) there was I think a time when -m 1 was interpreted as memory=1byte and this resulted in samtools spamming the disk with millions of tiny temporary files for each chunk. Can be that I mix it up with another tool, but it does not hurt to be explicit. But yes, as Pierre says this behaviour is expected.

[quentin54520]

Thanks for your reply. Yes in the real command i put 1G but i forgot here sorry 🙂

Answer 1

[Pierre Lindenbaum]

yes, it's normal. when it's sorted by coordinate, some similar DNA sequences are grouped in the same block of gzip compression, which improve the performance of the compression. When sorting by query-name, you break those groups.

see Size of BAM file reduces after sorting with samtools