Merge bam files to create a consensus?

[bwubb]

Would there be a way to create a consensus bam from multiple input bam files without skewing coverage to a ridiculous size?

Say I had several bams from exome sequencing of the exact same sample. It is my understanding that if I merge these files my overall coverage would be really high. Instead I might want to merge them but average the coverage across the bams rather then add them, however I can not think of a way to merge bam files in this way.

Thanks.

[dnusol]

I'm probably going to write something wrong, but if you had to perform several lanes/runs/experiments for one single exome, that means that you could not get sufficient coverage across all the exome in one single lane/run/experiment. So if you merge the bams your coverage will not be that large. As I understand it, coverage for one bam will be "randomly" dispersed, so merging the different bams will only make your coverage homogeneous and will cover the exome completely.

HTH

[volks]

Would there be a way to create a consensus bam from multiple input bam files without skewing coverage to a ridiculous size?

Say I had several bams from exome sequencing of the exact same sample. It is my understanding that if I merge these files my overall coverage would be really high. Instead I might want to merge them but average the coverage across the bams rather then add them, however I can not think of a way to merge bam files in this way.

Thanks.

please define ridiculous size. for whole-genome sequencing of human samples you can easily end up with alignments >100Gb.

[bwubb]

Beyond merging the same sample from multiple lanes/experiments...

I was playing around with the idea of quickly/efficiently creating a "baseline" bam file, so that the coverage present represents the average coverage across all files.

For instance, say I wanted to merge every exome.bam for CEO population samples of the 1000G data set into a single file. I could not think of a way to merge all of those files while averaging read depth or coverage instead of combining it.

In the end though I think I wanted to use this created file in a way that would not be good science.