search

rajanil / fastStructure

A variational framework for inferring population structure from SNP genotype data.
MIT License
133 stars 51 forks source link

color assignment not following K #71

Open rturba opened 2 years ago

rturba commented 2 years ago

Hello! It seems I'm getting a strange output with my analysis. When I run the distruct.py script on my dataset, the color scheme of the assignment plots do not seem to follow the number of K clusters. Has someone else seen that before? My best K is 2, and up to 3 the colors seem OK, but then it does not matter the number of Ks, I can only distinguish at most 4 different color patterns :(

I'm running this on a Linux cluster. I've processed my plink files to prune for LD and only retain biallelic SNPs.

rturba commented 2 years ago

I've ran it on the test data and is also behaving the same strange way... :/

k=3 image

k=8 image

apalmer828 commented 1 year ago

I've been seeing a similar thing, where when I make a distruct plot for k=3 I get two colors and at k=5 get three colors, etc. I've noticed it occurs with data from seeds but not others. I'm also running on a linux cluster .