Given a RNA-Seq data of a progeny between two inbred lines (for example an F4 plant which was derived from two inbred lines (parents) and the F1 selfed 3 times; however this is probably not crucial for the question).
We know the progeny should have only a few breakpoints in each chromosome, so one should expect long stretches (>1MB) with those 3 states:
- homozygous for parent1
- homozygous for parent2
In each state, most containing genes should therefore be reflecting this state, i.e. only expressing the parent1 allele in state 1) (i.e all SNPs support parent1) etc. A caveat would be uni-allelic expression at a heterozygous state 3), but this surely will not be the case for all genes in a given block (100+ genes per block are expected).
Both parental genomes and RNA-Seq libraries for both parents and F4 are available.
Naively I'd do SNP-calling (e.g.
GATK HaplotypeCaller or
bcftools call) of an F4 mapped against one parental reference and somehow combine the allele frequency in a sliding window and evaluate the likelihood of the given window to be in state 1-3. E.g. if all alleles support parent 1, it's probably state 1).
But I feel this must have been done before but can't quite find the right software or workflow software to use. But maybe I'm missing something. Ideally, a workflow would have as input the F4 RNA-Seq data (possibly alos parental RNA-Seq data), the parental genome(s) and as output a list of segments per chromosome.
I know, we could do gDNA sequencing for this even better, but I think RNA-Seq might also do the job and might save effort since already available.