Date: Friday, April 25th, 2025
Time: 10-11 am
Room: 68-181
CSB PhD Candidate: Kamal Maher
Advisor: Xiao Wang (Chemistry, Broad)
TDC Members: Peter Reddien (chair) Alex Shalek, Fabian Theis (external)
Title: Transcriptional harmonics of tissues
Abstract:
The structure and function of biological tissues are determined by interactions between cells. Spatial omics technologies enable measurement of the spatially constrained molecular features underlying such interactions. Several computational methods have been developed for unsupervised detection of condition-specific multicellular regions. However, there is no consistent quantitative definition of such regions, let alone the interactions that they are ultimately intended to represent. In this thesis, we provide such a definition using the language of harmonic analysis on graphs. Regions are defined in terms of positively}covarying (i.e. overlapping) low-frequency (i.e. large-scale) gene expression patterns over the tissue. This leads naturally to a definition of interactions in terms of negatively covarying (complementary) high-frequency (small-scale) patterns. Combinations of lows and highs (i.e. mids) can be interpreted as molecularly defined boundaries between regions. The fundamental nature of this quantitative framework reveals additional relationships between regions and interactions and further enables investigation in frequency space and at the subcellular scale. We provide demonstrations in simulations as well as datasets from multiple different experimental technologies, species, tissues, and diseases. Altogether, this work provides a rigorous quantitative framework within which current methods can be expressed and upon which future methods can be constructed.