Visual analysis of spatial transcriptomics data with RedeViz

Dehe Wang,Xianwen Ren

DOI: https://doi.org/10.1101/2024.07.09.602652

2024-07-13

Abstract:Spatial transcriptomics (ST) technologies are powerful tools to illustrate the spatial hierarchy and heterogeneity of tissues with the lens of multiplexed gene readouts. However, ST technologies generate sequence data rather than images, preventing intuitive examination of the cellular contexture of tissues. Moreover, the inherent sparsity of ST data caused by molecular crowdedness and sequencing dropouts poses great challenges to accurate and clear visualization. In this study, we introduce RedeViz, a toolkit crafted for enhancing and visualizing subcellular-resolution ST data. RedeViz applies a pixel-level enhancement strategy, visualizes ST data in automatic or custom-ized manners, and can display the cellular and genic spatial patterns with effects akin to HE staining. Strict evalua-tions confirm that RedeViz fits a wide range of ST platforms, including Xenium, Visium HD, MERFISH, CosMx, Stereo-seq, etc. The impressive performance of RedeViz across various scales from cell-, tissue-, organ-, to organ-ism-levels brings us a universal "What You See Is What You Get" framework for visual analysis of ST data.

Bioinformatics

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is how to enhance and visualize Spatial Transcriptomics (ST) data in order to achieve an intuitive presentation of high - resolution cell and gene spatial patterns. Specifically, the paper addresses the following key issues:

1. **Inherent sparsity of ST data**: Due to molecular crowding and sequencing loss, ST data are usually sparse, which makes it difficult to accurately and clearly visualize these data. Traditional visualization methods rely on manual annotation of cell types and color assignment, and are prone to overlook important cell structures, leading to potential inaccuracies or problems introduced by manual operations.

2. **Lack of intuitive cell structure presentation**: ST technology generates sequence data rather than images, which hinders the intuitive examination of tissue cell structures. Therefore, a method is needed to be able to transform these sequence data into an effect similar to that of HE - stained or immunofluorescence images, so as to more intuitively understand the spatial distribution of cells and genes.

3. **Data compatibility of different platforms**: Existing ST technologies include imaging - based methods (such as MERFISH, CosMx, Xenium) and next - generation sequencing (NGS) - based methods (such as Visium HD, Stereo - seq), and each method generates data with different formats and characteristics. A universal tool is required to process ST data from different platforms and provide consistent and high - quality visualization results.

To solve the above problems, the authors developed a toolkit named RedeViz. RedeViz enhances and visualizes ST data in the following ways:

- **Pixel - level enhancement strategy**: RedeViz applies a non - segmented pixel - level enhancement strategy, filling in the gaps caused by data sparsity and supplementing missing information by adding additional derived dimensions (dDims).

- **Automatic or custom - made visualization**: RedeViz can visualize ST data automatically or in a user - customized manner, presenting the spatial patterns of cells and genes with an effect similar to that of HE - stained.

- **Multi - scale visualization**: RedeViz supports fine - grained (pixel resolution) and coarse - grained visualization, being able to show both global features and retain high - resolution details.

Through these methods, RedeViz can provide a "what - you - see - is - what - you - get" (WYSIWYG) visualization effect at different scales (from cells to tissues, to organs and even the entire organism), thereby better understanding and analyzing spatial transcriptomics data.