Hi there, 

I have a few questions surrounding how the pedigree information works... I have 13 samples: mom, dad, and 11 progeny. The organism is a plant where sex is irrelevant. 

1. (How does / Does) incorporating a pedigree file in the process of running HaplotypeCaller on each sample (separately) and later when running GenotypeGVCFs on a combined gvcf actually improve the QUAL values? 

2. Is it possible to obtain genotypes in the progeny that make no sense in regards to Mendelian segregation, even when you incorporate the pedigree file during these mentioned steps?

I tried to use FindMendelianViolations to find out how often the calls violated possible segregation... but I have a bit of a complicated situation, and my progeny have different ploidies. It works out that each genotype can obtain 0, 1, or 2 alleles from the parents (depending on if it's mom or dad). It's because I'm working with a phased portion of an allopolyploid genome (only one of the subgenomes).

For example, if I have a triploid child, and mom's and dad's genotypes are:

Mom: A

Dad: A/T

You'd expect the child to be A/A/T (yes, the parents are different ploidy too.. *insert weeping here*)

But I'm sometimes seeing the call be erroneously reported as A/T/T. To make matters worse... sometimes an allele from Mom won't be inherited.

So, for a diploid child:

Mom: T

Dad: A/A

it could be T/A or A/A. 

Here's another example for diploid:

Mom: C

Dad: CC/TT

the child could be C/CC, C/TT, or CC/TT. 

Yet, sometimes I see the result being CC/CC. (this would have to have happened via double reduction... see note below about recombination)

I'm pretty confident there has been no recombination in this genomic region because of a previous QTL study done on this population. I should also mention that my progeny have much shallower depth than my parents - sometimes only a couple of reads support a site. I was hoping that my parental genotypes would remedy the depth issue of the children. But I see the problems described above more often than I'd like. 

Will the FindMendelianViolations tool not work for this situation because of the odd ploidy situation (admittedly I tried to run it and there was no output file...)? I think if I could get some clarity on my questions 1. and 2., I might have more confidence whether the pedigree information is being properly incorporated or not... 

Thanks so much for any insight / thoughts you might have. And if I'm unclear or confusing someplace, please let me know!

Kindly,

Charity