Adding read group to bam files from multiplexed samples

Question

[serpalma.v]

Hello

I have 60 samples (samp1...samp60), each one was barcoded and then pooled (10 samples/pool, 6 pools).

Each pool was sequenced in 9 lanes.

This leads to 1080 fastq files ( 60 samples * 9 lanes * 2 (PE) ) and 540 bam files.

I want to do variant calling with GATK.

I went through these two very informative posts:

https://gatkforums.broadinstitute.org/gatk/discussion/6472/read-groups

Read Group In Sam/Bam Files: What Do They Exactly Describe?

Accordingly, I am trying to define the read groups for each bam file, as follows.

• ID: flowcell ID and lane ID (i.e. HNTW5BBXX_1)
• SM: the name of the sample (i.e. samp31)
• PL: ILLUMINA
• LB: lib_samp31
• PI: insert size (i.e. 200)
• PU: flowcell ID and lane ID and sample ID (i.e. HNTW5BBXX_1_samp31)

I would like to clarify the following:

• Did I get something wrong interpreting the fields?
• Could I exclude PU?, as it is not required by GATK, according to the link above. Do you usually include it anyway?

Thanks in advance!

[swbarnes2]

Unless you have QC reasons to say that a lane did poorly, you should concatenate all 9 lanes together for each sample. Keeping them separate is doing you no favors. Merge the bams now before you do more.

[serpalma.v]

I read here that keeping bams separated during pre-processing is reasonable. And also, the way I understood it, for each sample, every bam file corresponds to a different read group, as they are derived from reads produced by different lanes.

[Devon Ryan]

5 year old recommendations are no longer relevant, just concatenate the lanes together.

[serpalma.v]

so then the read groups should be as follows:

• ID: samp31
• SM: samp31
• PL: ILLUMINA
• LB: samp31

Not sure about keepin PI and PU now...

Correct?