# tSNE

t-distributed stochastic neighbor embedding, or t-SNE, is a method for visualizing high-dimensional data, like flow cytometry data with many parameters, by mapping it to a 2 dimensional space. It's similiar to how a globe (3 dimensions) can be visualized with a map (2 dimensions). This can allow you to identify relationships easier, but can also distort the data (Africa is actually a lot bigger than you would think if you looked at a standard map of the world).

## Running tSNE

Unlike some cloud-based flow cytometry analysis platforms, the Floreada tSNE computation runs entirely on your own computer. We don't ever see your data.

You can run the tSNE algorithm by selecting the tSNE button from the toolbar. The popup menu will allow you to configure the run with several different options:

Iterations - tSNE works by running an algorithm over and over and over, improving the result each time. This represents the number of times you want it to run. Bigger values here generally result in better outputs, although there are very diminishing returns after some point. Bigger values here also are one of the most important factors in how long it takes tSNE to finish.

Learning Rate - Controls how much the weights are updated each time the algorithm runs

Perplexity - Optimal value depends on the density of your data. The more events you have, the bigger the value should be. 30 is a good default.

Mode - There are several flavors of the tSNE algorithm, of which only the Barnes Hut version is implemented in Floreada.

Parameters - Here you can select which parameters you want to use for the analysis. We would suggest, well compensated, fluorescent parameters which you are interested in. You must select at least 3.

Once you select "Run", tSNE will start. The process can take a long time, depending on how many events you have. If it is taking too long, try lowering the number of iterations or creating a gate to decrease the number of events. You can close the popup menu after starting and do other analysis while it runs in the background. You can check the progress of the tSNE run at anytime by again selecting tSNE from the toolbar.

## Cancelling tSNE

If you want to cancel a tSNE run, you can select the "Cancel" option from the popup menu. You will not get any data from a cancelled run.

## tSNE Results

After the tSNE run is complete, a "tSNE" population will appear under the population you ran it on. The plot for this population will have 2 parameters, "tSNE X" and "tSNE Y".

## tSNE Implementation Details

Floreada.io uses a modified version of this tSNE implementation, which is licensed under an MIT License. Floreada.io is *NOT* licensed under an MIT License. More Info.