spaDesign: A Statistical Framework to Improve the Design of Sequencing-based Spatial Transcriptomics Experiments

Juan Xie,Hyeongseon Jeon,Won Chang,Yeseul Jeon,Zihai Li,Qin Ma,Dongjun Chung

DOI: https://doi.org/10.1101/2024.09.10.612260

2024-09-14

Abstract:Rigorous experimental design is essential for obtaining biologically meaningful findings and high-throughput spatial transcriptomics studies are not an exception. While some efforts have been made to improve the design of these studies, it is still significantly understudied yet how to optimize key experimental factors of these experiments. In the case of sequencing-based spatial transcriptomics studies, determining the minimum sequencing depth is an important experimental factor to decide. To address this critical limitation, here we propose spaDesign, a statistical framework to improve the design of sequencing-based spatial transcriptomics experiments. spaDesign is a statistically rigorously designed framework that employs Poisson-Gaussian process and Fisher-Gaussian kernel mixture. It can easily simulate a range of spatial transcriptomics data with various sequencing depths, effect sizes, and spatial patterns, which allows rigorous estimation of needed total sequencing depths to detect spatial domains based on spatial transcriptomics experiments. We demonstrated the utility and power of spaDesign using 10X Visium data of the human brain and the chicken heart.

Bioinformatics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to optimize key experimental factors in high - throughput spatial transcriptomics experiments, especially determining the minimum sequencing depth required to detect the target spatial domains. Although there are currently some methods and techniques for improving the design of these experiments, how to optimize the key factors of these experiments is still under - studied. For sequencing - based spatial transcriptomics research, determining the minimum sequencing depth is an important experimental factor. To this end, the authors propose a statistical framework named spaDesign, which aims to estimate the total sequencing depth required to detect spatial domains by simulating data with different sequencing depths, effect sizes and spatial patterns, thereby improving the design quality of sequencing - based spatial transcriptomics experiments.