Entering edit mode
3.0 years ago
Anst
▴
50
Hello everyone!
I'm trying to project my ancient sample on PCA with 2 different basis: simons data and 1000. I am using smartpca from admixtools and eigenstrat files as input (exclude outliers). Something strange occurred, the ancient genome and its downsampling versions (cucrled) laid far away from each other and the main fork. Furthermore, if we use simons dataset as the basis the PCA plot looks different. Could you tell me why this can happen and how to fix this?
My guess is that you might have inconsistent alleles with the downsampled ancients, i.e., the reference and alternate alleles have been switched over. I would go back and check this in the original datasets by taking a SNP and making sure the alleles are the same.