SNP calling from pooled RNA-seq data

Question

[thjnant]

Hello,

I need to call SNPs from pooled RNA-seq data. In each pool, I have 2 or 3 individuals. I am following the GATK pipeline for this purpose.

What I wonder about, and this could be due to my lack of understanding of the SNP calling process from pooled data, if for a given position, one sees two alleles, how could one say whether this position is a heterozygote in one of the individuals or whether these two different alleles come from two homozygote individuals?

I would appreciate some comments on this. Thanks.

Answer 1

[Devon Ryan]

You actually can't call homo/heterozygote in cases like that, though you can determine if a position is a variant (well, part of the time at least). Even if GATK reports calls like that, you should ignore. In fact, this is a general problem with RNAseq where allele-specific expression can mask an underlying heterozygous haplotype.

[thjnant]

Thanks for your comment. From individually sequenced RNA-seq data is possible to see whether an individual is homozygote or heterozygote for a given position especially one can always check the bam files against the reference. But as you said, knowing whether a position is homozygote or heterozygote in pooled samples is impossible, the only information is that this position contains variation in a populations.

[Devon Ryan]

No, you can't always tell even in singly sequenced samples. You can call heterozygous sites in those, but not homozygous, since you could also be looking at allele-specific expression. That's among the many issues with using RNAseq for variant calling.

[thjnant]

Oh, yes, you are right. As one allele may not be expressed, so, then the truly heterozygote position will be called as a homozygote. Well, I should handle the limitations. Thanks again.

Answer 2

[Khader Shameer]

We recently published a method "eSNV-Detect" to call genotypes from RNA-seq data.

Publication here and software here.

Please give it a try and let us know if you have any questions