The Power of Spatial Omics for Unraveling Tissue Complexity - Seeker's Thoughts

Recent Posts

Seeker's Thoughts

A blog for the curious and the creative.

The Power of Spatial Omics for Unraveling Tissue Complexity


The Power of Spatial Omics for Unraveling Tissue Complexity







The Power of Spatial Omics for Unraveling Tissue Complexity


Spatial transcriptomics (or spatial RNA sequencing (sRNA-Seq), is an innovative way of mapping gene expression architecture within structurally preserved tissue sections. 


Learn how 10x Genomics' Visium platform and NanoString GeoMx Digital Spatial Profiler solutions empower scientists to link tissue morphology, biological activity, and gene expression using spatial genomics.


Identifying Cell Types


Spatial transcriptomics is an invaluable way of exploring cell dynamics in healthy and diseased tissues. By combining spatially resolved data with next generation sequencing technology and high-plex RNA technologies, spatial transcriptomics provides depth and clarity when it comes to understanding gene expression dynamics in specific niches.


Cell type identification is critical to understanding disease processes as well as discovering novel biomarkers and therapeutic targets. Unfortunately, manually labeling of tissue regions can be time consuming and error prone in densely packed tumor tissues; spatial omics technologies offer an efficient yet precise alternative solution.


Combining spatial transcriptomics with image analysis allows direct correlations between gene expression levels and tissue morphology and architecture, and image segmentation to identify cells of interest. Additionally, computational approaches have been devised that take advantage of spatial variation to model cell-cell interactions and isolate environmental effects on gene expression levels.


GeoMx DSP offers a global view of transcriptome while CosMx SMI delivers single cell and subcellular resolution for single RNA and protein measurements, creating spatial multiomics capabilities to analyze gene expression dynamics, TME interactions, cell-cell interactions and compare diseased with normal tissue samples for comparison purposes. Together these capabilities facilitate unbiased discovery that leads to more accurate clinical diagnosis and treatment decisions.


Identifying Intercellular Communication


Spatial multiomics allows researchers to track cells within tissues in order to discover intercellular communication pathways, providing a powerful complement to cutting-edge single cell technologies like RNA-seq or chromatin accessibility.


Spatial multiomics involves analyzing gene and protein expression at both an RNA or protein level in relation to their spatial context, thus providing key information about typical biological functioning as well as disease processes.


Analysis of such data requires specialized computational approaches. While technological innovation has resulted in an abundance of computational tools, navigating this vast landscape can be daunting. To prevent overfitting, best practice workflows must be established; otherwise an omics-based test developed within one laboratory might not perform as well when used elsewhere or even to the same population.


To address this challenge, we have recently developed spatial technologies combining laser capture microdissection with high-throughput omics profiling. With these tools we are able to profile tens of thousands of cells simultaneously within small sections of tissue by tracing back their location within it - this provides invaluable information about their biology within its ecosystem as well as insight into cellular interactions within its architecture.


Identifying Disease Signatures


As omics technologies develop, researchers are seeking ways to further interrogate the transcriptome and proteome of tissue samples. Spatial transcriptomics allows investigators to interrogate cell populations within their structurally preserved context and capture a snapshot of cellular architecture of tissue sections.


However, the spatial dimension of these datasets presents computational models with difficulties. Since feature measurements often exceed biologically meaningful signals by orders of magnitude, statistical algorithms may overfit. As detailed in MammaPrint case study, proper measures should be taken to avoid overfitting including fitting models on multiple datasets (Step 1) before verifying with independent sets (Step 2).


Spatial transcriptomics has the capacity to not only probe the transcriptome but also reveal disease signatures within tissues. Investigators can use preprocessing techniques like stLearn which normalize spots or segments according to their morphological similarity and nearby density using deep learning and histology image features (Step 1).


Investigators can use disease signatures to detect tumor-immune microenvironment patterns and paracrine interactions that could potentially contribute to disease processes, as has been done for triple-negative breast cancer, idiopathic pulmonary fibrosis, and esophageal squamous cell carcinoma among others.


Identifying Disease Progression


Like GPS tracks your position on Earth, spatial biology maps your omics data in 2D or 3D tissue contexts. Its powerful tools can help researchers decipher how cellular processes and intercellular communication play into disease progression as well as identify new biomarkers to guide improved clinical care.


As the initial step in computational model development, feature selection is essential. This involves selecting a set of measurements associated with clinical outcomes to reduce measurement burden and avoid overfitting (Tang et al, 2011; Teague et al, 2010).


Step two of feature validation involves testing model predictions on independent omics datasets that contain clinical outcomes similar to theirs, in order to assess whether or not it accurately predicts them and provide confidence in its predictions (Leek et al, 2010).


As part of this step, it is recommended that investigators utilize multiple omics datasets from multiple laboratories; models that perform well across these multiple datasets tend to perform better when applied clinically. A similar strategy can help avoid the possibility that an omics-based diagnostic test might fail for reasons unrelated to molecular characteristics measured (Tang et al, 2011; Paik et al, 2004). When creating spatial omics experiments, it is crucial to take into account both plex and resolution requirements. Hypothesis testing experiments would benefit from using imaging- and subcellular-sampling-based technologies like MERSCOPE, Esper and Xenium; while array-based technologies like Visium may be better suited for conducting atlas generation experiments.

No comments:

Post a Comment