Integration Analysis Of Single Cell And Spatial Transcriptome Data Based On G-S Mapping Algorithm

In 2020,Since it is being named “Method of the Year” by Nature Methods,spatial transcriptome sequencing technology has evolved rapidly over the past three years.In contrast to single-cell RNA sequencing(single-cell RNA sequencing,sc RNA-seq),spatial transcriptome sequencing technologies can obtain both gene expression and location information.Spatial information of the spatial transcriptome is crucial to our understanding of the inter-cellular communication behind normal and diseased tissues.However,due to technical limitations,classical spatial transcriptome data only can capture the transcripts of uniformly sized spatial dots.Because the cell size of different organs in different species varies greatly,it is difficult to achieve single-cell resolution in spatial transcriptome data.Therefore,the joint analysis of single-cell sequencing and spatial transcriptome sequencing data and realizing spatial transcriptome profiles at single-cell resolution,is of great importance to study the interactions of cell subpopulations in tissue development,regeneration and disease.To address the shortcomings of the technique and obtain spatial transcriptome data at single-cell resolution,we propose a spatial mapping-deconvolution algorithm at single-cell resolution based on the G-S stable matching while fusing both single-cell data and spatial transcriptome data.It provides a novel way to study the interactions among cells.It provides a brand-new way of revealing cellular interactions and studying the whole developmental process of tissues.The main work of this paper is as follows.First,we propose a novel algorithm DRIM(single-cell Deconvolution followed by Region-growing,Interpolation and iterative-Mapping)based on G-S algorithm,which infers the spatial mapping of biological tissues at full slice and single-cell resolution.It provides a clear and comprehensive map for studying the tissue development,regeneration and immune microenvironment of cells.Second,we simulated the low-resolution data by using Stereo-seq data,and then used DRIM to expand the spatial points of the low-resolution data to improve the resolution,proving the accuracy of the algorithm.We also implemented our new algorithm on real data of regeneration of vortex worms and mouse heart,combining their single-cell data and spatial transcriptome data to map the spatial atlas at singlecell resolution depending on the size of the cells.Third,a new algorithm Spatial Temporal Auto-Encoder(STAE)designed by combining a deep learning auto-encoder with the pseudo-time analysis on a spatiotemporal map at single-cell resolution to facilitate cellular appreciation,migration and differentiation during development.In summary,we firstly propose a novel algorithm for single-cell data and spatial transcriptome data fusion analysis.It realizes spatial transcriptome atlas reconstruction at single-cell resolution,and uses Stereo-seq data and real data to verify the accuracy of the algorithm.Finally,a new algorithm for cell migration differentiation analysis is designed for the spatiotemporal atlas at single-cell resolution,which provides a new way for the study of movement and change during cell development.