As explained in the “Visualizing Data using t-SNE” article, the aim of the tSNE algorithm is to visualize data. In particular - convert data of high dimension to low dimension (2 or 3 dimensions). In this project I will focus on 2-dimension conversion.
The tSNE algorithm takes into consideration only the relation between each point to the others. My goal is to emphasize this kind of behavior in the following way: reduce high dimensional data with tSNE to 2-dimensional space, but with a certain shape.
There is a tutorial code which can be used to see the behavior I mentioned above.
In the tutorial.ipynb file you can find a guidance on how to use the code, play around with it, and see the results.
my_t_sne.py is based on the work of Alexander Fabisch, Christopher Moody and Nick Travers, and is used under the BSD 3-clause license.
shapes.py and meshes.py are based on the work of Jonathan Viquerat, and is used under the MIT license.
tasic et al data and relatd files are based on files from https://portal.brain-map.org/atlases-and-data/rnaseq.
Special thanks to Ela Fallik and the Friedman Lab for their guidance and support throughout this project.