google / deepvariant

DeepVariant is an analysis pipeline that uses a deep neural network to call genetic variants from next-generation DNA sequencing data.
BSD 3-Clause "New" or "Revised" License
3.15k stars 709 forks source link

3-generation pedigree analysis possible? #594

Closed Solyris83 closed 1 year ago

Solyris83 commented 1 year ago

Hi, I am toying with the idea to get a 3-generation trio sample for a condition I am interested in and 1 parent and child generation has the condition while the grandparent generation is not. May I know if this kind of pedigree can be used for the tool effectively using the 3-generation pedigree structure or it can only take 2 generation?

AndrewCarroll commented 1 year ago

Hi @Solyris83

Although it is in theory possible to train a 3-generation trio, I'm not aware of a truth dataset with sufficient representation of grandparents-parents-child (n=4,n=2,n=1) that would allow us to train such a model. The platinum genomes pedigree around NA12878 is the closes, but I think that is missing high enough quality labels of the children and not all of the grandparents.

pichuan commented 1 year ago

Hi @Solyris83 , hopefully @AndrewCarroll 's answer have been helpful. I'm going to close this issue now. Feel free to let us know if you have more questions.