Mapping the Gene Space at Single-Cell Resolution with Gene Signal Pattern Analysis

Aarthi Venkat, Sam Leone, Scott Elliot Youlten, Hannah Thomas, Eric Fagerberg, John Attanasio, Nikhil S Joshi, Michael Perlmutter, Smita Krishnaswamy

Keywords: graph signal processing, manifold learning, single-cell sequencing, gene network analysis, gene embeddings

TL;DR: We present GSPA, a graph signal processing approach for single-cell and spatial transcriptomics which embeds gene expression features as signals on a cell-cell graph, enabling diverse cluster-independent and gene network-based downstream analysis.

Abstract: In single-cell and spatial sequencing analysis, several computational methods have been developed to map the cellular state space, but little has been done to map or create embeddings of the gene space. Here we formulate the gene embedding problem, design tasks with simulated single-cell data to evaluate representations, and establish relevant baselines. We then present gene signal pattern analysis (GSPA), a graph signal processing approach that learns rich gene representations from single-cell data using a dictionary of diffusion wavelets on the cell–cell graph. This approach embeds genes based on their patterning and localization on the cellular manifold, enabling characterization of genes for diverse biological tasks, including identifying gene coexpression modules and gene subnetworks associated with patient phenotypes.

Submission Type: Extended abstract (max 4 main pages).

Submission Number: 109