Analysis And Processing Of Single-Cell RNA Sequencing Data

With the development of science and technology,single-cell RNA sequencing(scRNAseq)has been widely used in the study of cellular heterogeneity,rare cell type inference,tissue heterogeneity,early embryo characterization,and neuroscience.However,due to the high sparseness,high dimensionality,and high noise of sequencing data,the analysis and processing of single-cell RNA sequencing data are subject to certain challenges in terms of accuracy and computational efficiency.With the development of deep learning,the analysis and processing of single-cell RNA sequencing data based on deep learning has increasingly become a research hotspot in bioinformatics.This topic analyzes and processes scRNA-seq data based on deep learning technology.The main research contents are as follows:A ZINB loss-based auto-encoder and KL loss-based deep embedded clustering scRNAseq data analysis and processing method(Based on Auto-encoder and KL loss Clustering)are proposed,called scAKC.The autoencoder based on the ZINB model maps scRNA-seq data to a nonlinear function of low-dimensional latent representation,adding random noise to the autoencoder,making it more powerful for data feature representation.Based on experiments on a large number of simulated data and real datasets,scAKC outperforms several other popular clustering algorithms in control experiments under various clustering performance metrics,and the running time increases linearly with the sample size.Accuracy and efficiency make scAKC a promising algorithm for analyzing large-scale scRNA-seq data.A scRNA-seq data analysis and processing method based on graph attention autoencoder and self-optimizing clustering(Based on Graph Attention Autoencoder and Self-Optimizing Clustering)is proposed,which is called scGSC.First construct a cell map and refine it by network denoising.Cluster-friendly representations of cells are then learned through an autoencoder,which propagates information among cells with different weights and captures latent relationships between cells.Finally,a self-optimizing clustering method was used to obtain cell clusters.Through a large number of simulated data and real data experiments,it is found that the clustering effect of scGSC is better than the current popular single-cell clustering algorithms,and it has good scalability on large-scale scRNA-seq datasets,which has important practical application value.