tSNE
What is tSNE?
Just a way to visualize high-dimensional data. In the case of single cell sequencing, we have say 10,000 cells, each with say expression of 25,000 transcripts. So that’s a 10,000 X 25,000 matrix, which we can’t plot in a 2D graph.
That basic problem can be solved by different methods, like Principal Component Analysis (PCA) or diffusion maps (DM).
tSNE is different because:
(1) It is an iterative machine learning algorithm, so it can produce different results each time it runs
(2) Rather than just plotting the points, tSNE explicitly tries to group points, making it a popular choice for single-cell group classifications.
Videos!
Last updated
