Mapping the Topography of Spatial Gene Expression with Interpretable Deep Learning

Uthsav Chitra, Brian J. Arnold, Hirak Sarkar, Cong Ma, Sereno Lopez-Darwin, Kohei Sanno, Benjamin J. Raphael

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Scopus citations

Abstract

Spatially resolved transcriptomics technologies provide high-throughput measurements of gene expression in a tissue slice, but the sparsity of this data complicates the analysis of spatial gene expression patterns. We address this issue by deriving a topographic map of a tissue slice—analogous to a map of elevation in a landscape—using a novel quantity called the isodepth. Contours of constant isodepth enclose spatial domains with distinct cell type composition, while gradients of the isodepth indicate spatial directions of maximum change in gene expression. We develop GASTON, an unsupervised and interpretable deep learning algorithm that simultaneously learns the isodepth, spatial gene expression gradients, and piecewise linear functions of the isodepth. GASTON models both continuous gradients and discontinuous spatial variation in the expression of individual genes. We show that GASTON accurately identifies spatial domains and marker genes in multiple SRT datasets.

Original languageEnglish (US)
Title of host publicationResearch in Computational Molecular Biology - 28th Annual International Conference, RECOMB 2024, Proceedings
EditorsJian Ma
PublisherSpringer Science and Business Media Deutschland GmbH
Pages368-371
Number of pages4
ISBN (Print)9781071639887
DOIs
StatePublished - 2024
Externally publishedYes
Event28th International Conference on Research in Computational Molecular Biology, RECOMB 2024 - Cambridge, United States
Duration: Apr 29 2024May 2 2024

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume14758 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference28th International Conference on Research in Computational Molecular Biology, RECOMB 2024
Country/TerritoryUnited States
CityCambridge
Period4/29/245/2/24

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • General Computer Science

Keywords

  • Spatial transcriptomics
  • deep learning
  • expression gradients
  • gene expression topography

Fingerprint

Dive into the research topics of 'Mapping the Topography of Spatial Gene Expression with Interpretable Deep Learning'. Together they form a unique fingerprint.

Cite this