samtools merge

[bair]

Hello all,

I have >150 sorted alignment bam files (>50G) by applying bwa, do I have to merge them into one single file before any further analysis?

If I just run SNP calling on each of the bam file and merge them together, how about the duplicates? It might be different with the result coming from a big single bam file.

Thanks in advance for any help.

[drio]

Hello all,

I have >150 sorted alignment bam files (>50G) by applying bwa, do I have to merge them into one single file before any further analysis?

If I just run SNP calling on each of the bam file and merge them together, how about the duplicates? It might be different with the result coming from a big single bam file.

Thanks in advance for any help.

1. Merge all the BAMs that belong to the __same library__.
2. Mark dups on those
3. Remove dups
4. Run your snp pipeline per each merged BAM.

[quinlana]

And be sure to use Picard's duplicate marker. It is superior to the one in samtools (the samtools docs suggest this as well).

Fear the duplicate...

[drio]

Right.. not only that. It is not recommended by the samtools authors.

From the samtools page:

"Samtools’ rmdup does not work for single-end data and does not remove duplicates across chromosomes. Picard is better."

I think it would be useful to explain this in the samtools rmdup help.

[bair]

Thank you guys.

Why just merge BAM in the same library? Indel/Snps may have different depth in different library, if we do not merge them into one big BAM file, how to setup filter for snp calling?