Closed klaramagdalena closed 2 years ago
Hi Klara,
Sorry that we haven't implemented this in the Seurat function yet -
If you want a workaround, we use the class:knn
function for prediction under the hood (see code here), and get the prediction confidence scores from setting prob = TRUE
. You could do the same, just making sure to use the harmonyPCs as the predictors for k-NN. Also note that that particular function is stochastic when there are ties, so you should set a seed.
Joyce
Hi Joyce,
thanks a lot, it worked! :)
Klara
Hi, thanks for the really nice tool!
I am running the seurat adjusted workflow and wanted saw that the argument 'confidence = TRUE' does not work for the knnPredict.Seurat() function.
Does anyone now another way of getting to this information? Thanks a lot!
Best, Klara