Hi Jbisanz,
Thanks for updating r code for DEICODE. Great help to R users.
I’m confused about the way the top 8 features were selected.
mutate(a=sqrt(PC1^2+PC2^2)) %>% top_n(8, a)
In the DEICODE tutorial, it wrote
*The important features with regard to sample clusters are not a single arrow but by the log ratio between features represented by arrows pointing in different directions. *
Could you explain a little bit more on this?
Another problem I’m struggling with is the heatmap drawing as the paper by Cameron. In Fig5A, rownames are sample names; colnames are features; values are feature loadings. But how can we match the sample name to feature names with loading?\
Best,
Yun