brentp/peddy

PCA projection - strange population classification

numicator opened this issue · 3 comments

numicator commented

Hi,

I am analyzing a kindred - two parents and two kids. The PCA projection of the samples is attached below. Mother is the sample of clearly EU ancestry, PC2 ~ -1.5, Father is in my opinion of clear SAS ancestry, PC2 ~ -0.4, The children are the two dots in between. As you can see from the PC1/PC2 plot, the father looks pretty much SAS. Peddy classified all samples as of EU ancestry. Why is it so?

aadry_cohort0001 pca_check

brentp commented

hmm, that is an odd one. this can happen some times. there is a 4th dimension, not plotted that the svm uses to cluster as well. i will be working on improving ancestry classification over the next few weeks in somalier -- but that is not accurate yet, for cases like this one.

numicator commented

yap, it's a strange one. Seems individual of mixed ancestry. BTW, are you planning further active development of peddy or shifting to somalier?

brentp commented

peddy is stable, but I will continue to fix any bugs that arise. other than that, I will focus more on somalier.