(What to make instead of) UMAP plots (for genomics data)

Dimensionality reduction is commonly applied as a first step prior to analysis of data in a variety of disciplines. In machine learning, dimensionality reduction often is the analysis via embedding of data in a “latent space”. In the field of single-cell genomics, dimensionality reduction is particularly popular, specifically dimensionality reduction to two dimensions. After discussing the motivation for dimensionality reduction in single-cell genomics, I will present several results, along with open problems, related to how and when one should perform dimensionality reduction, with a focus on insights gleaned from the genomics field.

Bio

https://www.bbe.caltech.edu/people/lior-s-pachter?back_url=%2Fpeople