imputation through beagle

Question

[analyst]

Hi scientists,

I have 80 samples of wheat GBS data. I have called variants through GATK pipeline. Now I have to perform imputation.

Do I need to use these 80 samples for building reference panel?

Also please let me know if reference panel of wheat is available for imputation. I will use that if applicable.

Thanks for your help!

Answer 1

[dthorbur]

Building a reference panel from the samples you intend to impute seems like a very circular approach. Plus, GBS is a reduced representation approach, which comes with it's own problems that wouldn't produce a good panel anyway.

Depending on the variety you are using, you can look into Plant-ImputeDB, which has a wheat reference panel.

If not, reference panels are optional in some tools. Though, the more gaps you have in coverage, the lower your imputation success will be. Hence, GBS data without a panel will likely perform quite poorly even in the best case scenarios.

EDIT: I just noticed your other similar post about imputation. It seems pretty clear you need to do some more reading about imputation, how it works, what reference panels are, and what are the limitations based on different data types. This would go a long way in helping you move forward.

[analyst]

Thank you so much dthorbur. Yes I am new to this kind of analysis. I will follow your kind suggestions.

Thanks again!