bcftools mpileup sample labels, seg fault

Question

[Gregor Rot]

Running this kind of command gives seg fault:

bcftools mpileup -s sample1,sample2 -f genome.fasta sample1.bam sample2.bam

However this works (without -s):

bcftools mpileup -f genome.fasta sample1.bam sample2.bam

bcftools 1.7
Using htslib 1.7-2

Any ideas why this could be?

Thanks,
Gregor

Answer 1

[Pierre Lindenbaum]

1. bcftools 1.7 was released 3 years ago. You should upgrade it.

2. did you set the read-groups ? what is the output of

   samtools view -H sample1.bam | grep '@RG'
   

[Gregor Rot]

Thank you Pierre, silly me, updated to the latest version now.

There is no more seg fault, however I get this now:

[mpileup] failed to find a file header with usable read groups

And like you wrote, there is no flag @RG in the bam files. Can you help me out, what is the easies way to set this in the bam files? I suppose I can set each bam file with it's unique RG? Thanks

[Pierre Lindenbaum]

Adding Read Groups To Bam Files

[Gregor Rot]

Thanks for all the help, however I still get the same error after adding RG to bam files.

bcftools mpileup -s sample1,sample2 -f genome.fasta sample1.bam sample2.bam

If I do samtools view -H sample1.bam | grep "@RG" I get: @RG ID:sample1, same for sample2. So my bam files have RG.

Any ideas on why this still doesn't work? If I ommit the -s sample1,sample2 everything works.

bcftools 1.12
Using htslib 1.12

Thanks, Gregor

[Pierre Lindenbaum]

I get: @RG ID:sample1,

and you should have

@RG ID:whatever_for_sample1 SN:sample1

[Gregor Rot]

Thanks, and haha, this will take another over-night job to add this to all the bam files I have :-)

[Gregor Rot]

Can I ask, does bctools consider the @RG SN flag from bam files if present? If yes, does specifying the -s sample1,sample2 override the info from the bam files? But then again, why it doesn't work with the -s if bam files don't have the RG field. A little bit confusing for me, Thanks

[Pierre Lindenbaum]

the sample name is defined in RG/SN not RG/id, see the sam specification.

[Gregor Rot]

Thanks Pierre yes I understand that, however does the bcftools -s set the sample names for VCF output or just check if they match the ones defined in bam RG:SN? I just find it confusing, intuitively -s would probably set them, and if not, why doesn't bcftools just read them from the bam files? And why with -s the bam files would also need an RG:SN if -s would set them. Confusing to me.

[Pierre Lindenbaum]

the option will only use the samples defined with -s. You could have only one bam with 1000 samples. -s "S1,S2" would use only the samples S1 and S2.

Answer 2

[Gregor Rot]

Me again, so now I understood that bcftools mpileup -s sample1,sample2 will only consider sample1 and sample2 from the marked alignments (read in from bam files). Just as an exercise, I added sample1 and sample2 to the @RG SN of 2 bam files:

samtools view -H sample1.bam | grep "@RG"
# @RG     ID:sample1    SN:sample1

samtools view -H sample2.bam | grep "@RG"
# @RG     ID:sample2    SN:sample2

However, if now I do:

bcftools mpileup -s sample1,sample2 -f genome.fasta sample1.bam sample2.bam

I still get this error:

[mpileup] failed to find a file header with usable read groups

Please advise, thanks, Gregor

[Pierre Lindenbaum]

works on my machine:

bcftools mpileup  -s S1,S2 --fasta-ref ~/src/jvarkit/src/test/resources/rotavirus_rf.fa ~/src/jvarkit/src/test/resources/S1.bam ~/src/jvarkit/src/test/resources/S2.bam | grep CHROM

CHROM POS ID REF ALT QUAL FILTER INFO FORMAT S1 S2

$ for F in ~/src/jvarkit/src/test/resources/S1.bam ~/src/jvarkit/src/test/resources/S2.bam ; do echo $F && samtools view -H $F | grep '^@RG' ; done /home/lindenbaum-p/src/jvarkit/src/test/resources/S1.bam @RG ID:S1 SM:S1 LB:L1 CN:Nantes /home/lindenbaum-p/src/jvarkit/src/test/resources/S2.bam @RG ID:S2 SM:S2 LB:L2 CN:Nantes ```

Ah, my bad, I wrote SN, but it's SM; How did you add the read group ? by hand ?

[Gregor Rot]

Aaaa, it's that yes, SM, I added the RG flag with this:

samtools addreplacerg -r ID:sample1 -r SN:sample1 sample1.bam -o sample1_RG.bam

No problem, will change SN to SM and it should work. Thanks Pierre!