Skip to main content
To KTH's start page To KTH's start page

Computational methods for analysis and visualization of spatially resolved transcriptomes

Time: Fri 2019-11-15 10.00

Location: Air and Fire, Tomtebodavägen 23a, Solna (English)

Subject area: Biotechnology

Doctoral student: Jose Fernandez Navarro , Genteknologi, Science for Life Laboratory, SciLifeLab

Opponent: PhD Martin Hemberg,

Supervisor: Professor Joakim Lundeberg, Science for Life Laboratory, SciLifeLab, Genteknologi

Export to calendar


Characterizing the expression level of genes (transcriptome) in cells and tis- sues is essential for understanding the biological processes of multicellular or- ganisms. RNA sequencing (RNA-seq) has gained traction in the last decade as a powerful tool that provides an accurate quantitative representation of the transcriptome in tissues. RNA-seq methods are, however, limited by the fact that they provide an average representation of the transcriptome across the tissue. Single cell RNA sequencing (scRNA-seq) provides quantitative gene expression levels of individual cells. This enables the molecular characteri- zation of cell types in health, disease and developmental tissues. However, scRNA-seq lacks the spatial context needed to understand how cells interact and their microenvironment. Current methods that provide spatially resolved gene expression levels are limited by a low throughput and the fact that the target genes must be known in advance.

Spatial Transcriptomics (ST) is a novel method that combines high-resolution imaging with high-throughput sequencing. ST provides spatially resolved gene expression levels in tissue sections. The first part of the work presented in this thesis (Papers I, II, III and IV) revolves around the ST method and the development of the computational tools required to process, analyse and visualize ST data.

Furthermore, the ST method was utilized to construct a three-dimensional (3D) molecular atlas of the adult mouse brain using 75 consecutive coronal sections (Paper V). We show that the molecular clusters obtained by unsu- pervised clustering of the atlas highly correlates with the Allen Brain Atlas. The molecular clusters provide new insights in the organization of regions like the hippocampus or the amygdala. We show that the molecular atlas can be used to spatially map single cells (scRNA-seq) onto the clusters and that only a handful of genes is required to define the brain regions at a molecular level.

Finally, the hippocampus and the olfactory bulb of transgenic mice mim- icking the Alzheimer’s disease (AD) were spatially characterized using the ST method (Paper VI). Dierential expression analysis revealed genes central in areas highly cited as important in AD including lipid metabolism, cellular bioenergetics, mitochondrial function, stress response and neurotransmission.