Finally got to reading this! Super interesting & informative! 2 Qs: 1) I always do scaling for PCA even when it's all from a single assay type (indeed that's when I usually do PCA) eg microarrays for transcriptomics or LC-MS for metabolomics. I guess I should stop doing that?
-
-
-
I don't quite know what you mean by scaling: rescaling the variances of each of the variables to 1 is fine, but pretending that the PCs all scale to have the same variance is wrong, since by definition the variance of the first one is larger than the second.
- Još 10 drugih odgovora
Novi razgovor -
-
-
Don't want to be nit-picky, but the ggplot2 ratio setting function is called coord_fixed(), not coords_fixed(). But a very informative read, thank you!
-
Sorry, the original paper was in latex, when the editors transposed to word, there were many things that changed..I am sure we missed other things, but the rmd must have the write syntax right ?
Kraj razgovora
Novi razgovor -
-
-
Definitely covering this paper for our next data journal club meeting! Fantastic read and very informative and educational!
-
Thanks..
Kraj razgovora
Novi razgovor -
-
-
Nice! Is "10 quick tips" the new "10 simple rules"?
-
They told us the rules for management, career etc and tips would be better for scientific recommendations.
- Još 1 odgovor
Novi razgovor -
-
-
I have one question on tip 8: In the cytometry community, I think it is very common to perform density-based clustering on tSNE results (or viSNE, HSNE). What do you think about these practices, then? Thank you for your time.
-
I think it is a double wammy: tsne clusters the data then you cluster again, but denisty based clustering is not appropriate because tsne does not respect the original data densities, see
@nlanhuong thesis which should be out quite soon. - Još 1 odgovor
Novi razgovor -
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.
