Study On Cell Differntiation Trajectory Algorithm Based On Single Cell Sequencing Data

The development of single-cell sequencing technology allows us to obtain large amounts of single-cell transcriptome data at low cost and with high precision,making it a powerful tool for revealing intercellular heterogeneity and prying cells.When analyzing the complexity of gene dynamics and variability in cell development,reconstruct the pseudo-time trajectory of cell differentiation with single-cell gene expression data remains a hot topic as well as a challenge.Although many computational and statistical methods for single-cell analysis have been developed recently,existing methods have their own limitations,so more efficient and accurate algorithms are needed to be proposed.Reconstruction the pseudo-time trajectory of cell differentiation contains two parts,which are data processing and cell sorting.Among them,data processing is the basis of algorithm,which ensures high accurancy of cell sorting.Single-cell sequencing data used in this paper is a large-dimensional sparse matrix,whose gene expression data has a span of magnitude,which increases the difficulity of analysis and challenges the applicability of our algorithm.In order to solve this problem,we need to clean,normalize our data,and then reduce its size.Firstly,this paper uses dpFeature to select useful genes by deleting genes that have little effect on differentiation process.In order to eliminate the magnitude of the data,we need to normalize our data logarithmically.Finally,based on the characteristics and structure of the genetic data,we choose the MLLE dimensionality reduction method to reduce the size of our data to eliminate data noise and facilitate data visualization.In the cell sorting part,this paper first proposes a dynamic radius neighbor method to find cell-marker and cluster cell clusters;then use cell-marker to construct the minimum spanning tree to describe the overall branching framework of cell differentiation trajectory;A new cell pseudo-time calculation method based on the Apollonius circle sorts the cells.In the verification part,the method was applied to the human embryonic stem cell dataset and the mouse embryo single cell dataset,and the prediction trajectory of the empirical results and the laboratory differentiation trajectory results were analyzed by Spearman correlation,the results showd 0.872 and 0.894,which are good result.