Closed huddlej closed 2 years ago
Pushed a change to fix this. The old code has been made "raw" for now, but I can delete it completely if we decide these graphs look better.
@nandsra21 Can you also use the same approach for 2018-2020 figure colors? Thank you!
We can reopen if needed - but for right now we're only making HDBSCAN figures so "colors" is fine.
The auto-assigned seasonal flu clade colors worked for a little while, but the latest iteration of the embeddings assigns colors that make related clades almost impossible to distinguish. Older clades that barely have any data get colors that should be used for larger extant clades.
For example, A2, A2/re, A3, and A4 are almost identically colored in the following figure:
We should go back to manually assigning colors to these clades, since there are a fixed number of clades in the analyses.