Variant calling for a polyploid species

Question

[User000]

Dear all,

I am working with a tetraploid species. I have RNA-seq for varieties and a reference genome. I am going to do variant calling, possibly to study intra-varietal and inter-homoeologous SNPs.

• Is it OK to do variant calling for a every variety separately is I have on average around 30K-50K reads (from read depth plotted over chromosome) ?
• Because I am not sure what analysis I can do in case of cohort (joint) variant calling.
• when doing variant calling how can I deal with the problem of homoeologous sequences? (A & B subgenomes ie.)? I am using -hisat2/tophat for alignment -GATK/FreeBayes for variant calling

Any help, a link to a publication etc is appreciated!

Answer 1

[Nicolas Rosewick]

You could start by looking at GATK best practices for variant calling from RNA-Seq :

https://github.com/gatk-workflows/gatk3-4-rnaseq-germline-snps-indels

In brief :

• Alignment with STAR
• SplitNCigarReads from GATK
• BQSR from GATK
• Haplotype caller GATK
• Some filtering

[User000]

Hi Nicolas thank you! I already did all of these steps. But I have very specific questions I stated above.

[Nicolas Rosewick]

Ok. Maybe you should edit your question by adding the analysis you already did.

[User000]

I did state the programs I am using. It is not so difficult to find the best practice etc, however I think it is not so straightforward when dealing with the polyploid species. So I wanted to know if there are some particular parameters to use or filtering etc. Thank you anyway

Answer 2

[Kritika]

Is it OK to do variant calling for a every variety separately is I have on average around 30K-50K reads (from read depth plotted over chromosome) ? it is fine to do varinat calling but my concern is how much depth you can select to filter it out the low depth reads as you have RNA seq data. Usually it always good to do WGS for Variant calling

when doing variant calling how can I deal with the problem of homoeologous sequences? (A & B subgenomes ie.)? I am using -hisat2/tophat for alignment -GATK/FreeBayes for variant calling Use BWA for SNP calling but you have RNA seq data i would say go with Hisat and freebayes for variant calling You classify your reference into A and B sub genome for example chromosome A1-A3 and B1-B3 total chromsome 6 in total so in case any homoelogous reads in mapping on A will also may have map on B so you will have clear picture of this.

[User000]

Hi thanks! My chromosomes are already divided in A and B, but when I do variant calling is it nto going to result in false calls? due to the fact that it will map both on A & b? How to deal with this?

[Kritika]

See this is reason i said for SNP calling use BWA or Bowtie because this tools will map the reads with high confidence and accuracy But you have RNA seq data so it is good to map with hisat because hisat and tophat will span exon and intronic region and map the reads in protein coding region.

Answer 3

[Omics data mining]

Hello

You can check SNiPloid https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3791807/

This article can help you to go further in order to predict homeologus SNPs .

I hope it will help you.