Samtools index/sort/rmdup

[hanleng]

I am trying to filter several steps before the mpileup:
1) samtools sort -n in.bam out.bam # sort the bam file
2) samtools rmdup -sS in.srt.bam out.bam # remove PCR duplicates
3) samtools view -b -F 1024 -q 20 *.bam > out.bam # filter duplicates reads and mapping quality
4) samools index .bam # give a index so that for mpileup

However, for the index step, it shows
"[bam_index_core] the alignment is not sorted: reads without coordinates prior to reads with coordinates." Anything wrong here?

[peromhc]

samtools sort -n sorts by name, not coordinate.. index requires sort by coordinate..

[hanleng]

samtools sort -n sorts by name, not coordinate.. index requires sort by coordinate..

So that I can sort without "-n" again and then index the bam file? Thanks.

[peromhc]

So that I can sort without "-n" again and then index the bam file? Thanks.

yep.. you should check out the git version with multithreaded sort -@ flag.. you're doing a lot of sorting!

[hanleng]

yep.. you should check out the git version with multithreaded sort -@ flag.. you're doing a lot of sorting!

I do not want to a lot of sorting - I just want to filter some reads before doing the mpileup...

[swbarnes2]

I am trying to filter several steps before the mpileup:
1) samtools sort -n in.bam out.bam # sort the bam file
2) samtools rmdup -sS in.srt.bam out.bam # remove PCR duplicates
3) samtools view -b -F 1024 -q 20 *.bam > out.bam # filter duplicates reads and mapping quality
4) samools index .bam # give a index so that for mpileup

samtools rmdup doesn't mark duplicates with 1024, it just removes them. So you don't need to filter for that. And why not filter before you sort? It would save a bit of time.

And mpileup doesn't actually require an index.

[hanleng]

samtools rmdup doesn't mark duplicates with 1024, it just removes them. So you don't need to filter for that. And why not filter before you sort? It would save a bit of time.

And mpileup doesn't actually require an index.

According to my understanding, rmdup remove PCR duplicates, and -F 1024 remove duplicates reads, right? I just want to remove all kinds of duplicates to call SNP, so is this correct?

[swbarnes2]

According to my understanding, rmdup remove PCR duplicates, and -F 1024 remove duplicates reads, right? I just want to remove all kinds of duplicates to call SNP, so is this correct?

-F 1024 will remove all reads that have the 1024 flag set. If you don't run any software which will ever set that flag, then there's nothing with that flag to remove. I don't think anything in the samtools suite will ever set that flag. Picard's MarkDuplicates will, but samtools rmdup won't.