Freebayes tumor vs. normal inquiry

Question

[alons]

HI all,

I've been trying to run Freebayes variant caller on a pair of tumor-normal samples but I can't seem to get the bar column for the normal bam file or the tumor. The commands I tried were:

 freebayes -F 0.01 -C 2 --pooled-continuous --fasta-reference efgenome.fa tumor/aligned_sorted.bam normal/aligned_sorted.bam > var_bayes_tumor_normal.vcf

 freebayes -F 0.01 -C 2 --pooled-continuous --fasta-reference efgenome.fa tumor/aligned_sorted.bam -b normal/aligned_sorted.bam > var_bayes_tumor_normal.vcf

 freebayes -F 0.01 -C 2 --pooled-continuous --fasta-reference efgenome.fa -b tumor/aligned_sorted.bam -b normal/aligned_sorted.bam > var_bayes_tumor_normal.vcf

Each of them only gave me one bar column, for example:

CHROM POS ID REF ALT QUAL FILTER INFO FORMAT bar

chr10 123240474 . G T 61.0498 . AB=0;ABP=0;AC=2... GT:DP:RO:QR:AO:QA:GL 1/1:2:0:0:2:77:-7.30883,-0.60206,0

I've read that Freebayes doesn't have a run option specifically for tumor-normal analysis but you can supply it with 2 bam files (or more) and do the analysis yourself. Has anyone succeeded in running freebayes on 2 bam files and getting 2 bar columns in the VCF or know how to interpret the VCF in regard to it's run on 2 separate bam files?

[shrutidesai0208]

Hi Herzliya, I am getting the same issue and looking out for solution. Did you get turn around this?

Answer 1

[sambrightman]

I believe this is caused by having the same ID field in the RG tag for the normal and tumour samples. Freebayes needs a unique sample ID, which I think is not always the case if samples were in the same run/lane. I reheader-ed by tumour BAM and got both columns in the output.

As far as I can tell the fact that Freebayes does this silently is a bug. It has a check for duplicate IDs and should exit with a message when they occur. However, it is ultimately relying on SamReadGroupDictionary for the combined headers, which silently discards read groups with the same ID before they reach Freebayes' check.