PCA results shows mixed population ?

Question

[putty]

Hello everyone, I ran PCA to check for population stratification before imputation. Looking at the plot it seems i have a mixed population (EUR+AMR). I plan to impute using Michigan imputation server for which i need to choose the population of my uploaded data. My confusions:

1) Do i choose a mixed population as my sample population based on my PCA plots. (This dataset was presumably recruited from an European cohort but the PCA doesn't show a clearcut EUR population but a mixture of EUR+AMR).

2) What are the possible consequences if i choose to impute as a EUR population (i don't like this idea)

Thank you all.

[4galaxy77]

Does it really make you choose the ethnicity of your samples? Or are you trying to decide which reference panel you use? Imputation doesn't change based on the ethnicity of the samples (i.e. there isn't an option in imputation algorithms where you specify the ethnicity of your samples). Your data seems to contain a few individuals which are slightly admixed, but are primiarily of EUR ancestry. In that case, I would just use the full HRC reference panel as it contains lots of EUR reference samples.

[putty]

Thank you for the response. Yup the server does ask you about population ethnicity (i am not sure whether this is for the reference panel or my own sample). First we are asked to select the reference panel and in the next section the population. But as you have suggested, majority of my samples appear to be clustering with the EUR so i might aswell go with the EUR population.

Thank you