search

epurdom / clusterExperiment

R package of techniques for comparing clusterings of single-cell sequencing data
36 stars 13 forks source link

dimReduce and scaling #213

Open epurdom opened 7 years ago

epurdom commented 7 years ago

If do not do PCA for dimReduce, should we have ability to scale the data before clustering? An option with dimReduce? Except you might want to both filter genes and work on scaled data.

What about when we do the hierarchy for dendrogram? Default doesn't use PCA, but just top variable genes, so not based on scaled data.

@drisso: thoughts?

drisso commented 7 years ago

What do you mean exactly by scaling the data?

epurdom commented 7 years ago

I mean per gene scaling, so the observed variance per gene is 1 for all genes; basically a scale() command. The idea that if you work not on dim. reduced data, but a set of filtered genes to cluster, you might want to have the data on all of the genes on the same scale (in the log space).

On Jul 19, 2017, at 7:24 AM, Davide Risso notifications@github.com wrote:

What do you mean exactly by scaling the data?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/epurdom/clusterExperiment/issues/213#issuecomment-316402320, or mute the thread https://github.com/notifications/unsubscribe-auth/AHXGVQVmzSkCWEtjOdoC7WISjKqtq9wMks5sPhBFgaJpZM4OLG80.