Graph representation and niches reconstruction - January 2025 (6 of 9)

Описание: This lecture introduces the concept of niches in spatial transcriptomics and discusses strategies for their computational representation using spatial graphs. Drawing from biological and mathematical principles, the session explores how spatial proximity and cellular context can be used to define functionally relevant microenvironments in tissue.

Topics include:

Biological definition of a niche vs. domain

The use of spatial graphs (e.g. radius- or KNN-based) for modelling cellular neighbourhoods

Transcript-based vs. cell-type-based niche representations

Aggregation strategies for generating niche-level expression profiles (e.g., mean, kernel-based weighting)

Discussion of batch effects and biological heterogeneity across tissues and patients

Applications of niche annotation in tumour microenvironment analysis, spatial metrics, and cell–cell communication studies

Introduction to the NicheFormer framework for mapping niche identities across modalities

Outlook on how spatial niches can inform prognosis and patient stratification

This session underscores the role of spatial context in uncovering cellular function and highlights how niche representations can enhance both exploratory and predictive spatial omics analyses.

The video was recorded live during the course “Spatial Omics Data Analysis” (https://elixir-europe-training.github...) on 23 January 2025 in Lausanne.

Target audience: PhD students, postdocs, and researchers who are involved (or will be in the near future) in projects including spatially-resolved transcriptomics (SRT) data and want to acquire the skills to get started with spatial data analysis.

Speaker: Anna Schaar, Helmholtz Zentrum München, Germany.

Any questions about this video? Contact us [email protected]