Join us this Wednesday for a talk by CSB's Brian Cleary of the Regev and Lander labs! It's at 12:30 in 16-220.
Title: Compressible representations of gene expression patterns reveal fundamental processes and enable highly efficient transcriptomics
Natural images are compressible because they are often highly structured -- for example, a photograph of a face might be structured by curved edges and fields, and, unlike a random collection of pixels, can be represented with a relatively small number of wavelet coefficients. For scenes that are known to be compressible, images can be efficiently captured directly in a compressed format, and then decompressed for viewing. I argue that the same is true for gene expression profiles; they are highly structured, compressible, and can be captured with high efficiency in a compressed format. Making compressed measurements can potentially save an order of magnitude or more in experimental costs. Moreover, finding compressible representations for gene expression reveals fundamental programs and pathways utilized by the cell.