charite
Finer granular inheritance modes
I think for the practitioner, the following additional inheritance modes would make sense:
DE_NOVO(either in addition or instead of AD/XD if one wants to screen for non-de novo AD/XD first and is not interested in the possible false-positive calls)COMPOSITE_RECESSIVE(either in addition to or instead of AR/XR if one wants to exclude this case in a first step)
For DE_NOVO -- basically, if we just make this token to be a synonym for AUTOSOMAL DOMINANT this would work in many cases. There are however also cases of dominant X-recessive diseases. It is possibly to prove that 1/3 of all cases of lethal X-recessive diseases must be de novo mutations, for instances. Do we currently call de novo X-chromosomal mutations? Thus, the mother is HOM REF and the son is hemiyzgous ALT? I think what Manuel is getting after with COMPOSITE RECESSIVE is that some practitioners look for compound het recessive mutations in outbred families. To me this doesntmake sense, because they can just use the recessive filter and if they do see a good homoyygous candidate then they shouldnt want to miss it, but this is indeed common practice in some circles, and so we might want to support it. Skype?
My only point was to make post-filtering VCF files easier by not having to look at genotypes (as this also involves looking at the pedigree) but just ask "is de novo in the list of compatible variants".
The same holds for composite heterozygous as this is not the first thing to look at in consangenous families.
If there were additional special markers in the VCF files, this could be done with grep alone.
I don't understand, do you want to write a label into the annotated VCF file? If people are grepping, there are likely to be just as many people who are looking for RECESSIVE and expecting to find both as vice versa. Also, there is the X chromosomal whichI do not think is implemented. Can we talk about this before you implement anything?
I am not sure this is a great idea, since there are many "gotchas" -- false positive or false negative variants calls are one of them. I think this should be a special tool rather than a library function -- but maybe I do not understand what we are talking about, sorry I could not take the call today, let's try again next week.
@pnrobinson let's exclude de novo for now but I think we can agree that an additional flag on whether the variant of interest is homozygous or heterozygous will not harm. I agree that people can always use another software or eyeball the variant but I'd like people to be able to grep out the compound heterozygous variants compatible with recessive mode of inheritance if they have a consanguineous case at hand.
#327 adds additional annotation on recessive annotation
The remaining missing thing for now is mitochondrial