Picard Addreplacereadgroup Error: Rg Id Not Found In Header

Question

[William]

I have a bam file which is missing read groups in the header.

Picard ValidateSamFile.jar gives : ERROR: Record 1, Read name M0VN6:402:2486, RG ID on SAMRecord not found in header: M0VN6 ...

So I try to use Picard AddOrReplaceReadGroups:

AddOrReplaceReadGroups.jar I= myBam.bam O= myBam.readGroupReplaced.bam SORT_ORDER=coordinate RGID=myBam RGLB=myBam RGPL=IONTORRENT RGSM=myBam RGPU=myBam

This gives the error:

ERROR: Record 1, Read name M0VN6:402:2486, RG ID on SAMRecord not found in header: M0VN6 ...

Why does Picard AddOrReplaceReadGroups crash about this? This is what I am trying to fix by using that function. Or do I misunderstand what AddOrReplaceReadGroups does?

What I get from the AddOrReplaceReadGroups discription is that it replaces all read groups in the Bam file with the one specified and assigns all reads to this read group.

[Matt Shirley]

You might want to have a look at the SAM format specifications

Answer 1

[Matt Shirley]

You have read group tags in each individual read (RGID). What you are telling Picard to do is add a conflicting @RG line to the header of your SAM file. You should just add the RG header line in the format:

@RG ID:id   PL:iontorrent   LB:library  SM:sample

This is all you need, since your reads already have RG ids. You should be able to verify this by looking at a read and finding the RG:Z:id tag.

[William]

Ok how do I ad a read group to the bam header? Do I manually copy and edit the header to a file and then use picard ReplaceSamHeader?

[Matt Shirley]

That would be fine. You can do the following to append the @RG line to your BAM header:

echo '@RG\tID:id\tPL:iontorrent\tLB:library\tSM:sample' > rg
samtools view -H my.bam | cat - rg | header
samtools reheader header my.bam

[eva]

i guess you are missing the '-e' option for echo, otherwise the tab wont be interpreted correctly.

[seidel]

There's a typo here, right? You don't want 'echo' there twice, and the read group fields have to be separated by a tab.

[Matt Shirley]

Yes, thanks. I am not sure what happened but it's fixed now.