Data Analysis Based On Dimension Reduction And Clustering Of Single Cell RNA Sequencing

Single cell RNA sequencing is the new generation sequencing technology.It not only has the advantages of high throughput and deep sequencing,but also can measure the properties of individual cell states accurately and decrease the correlation of each cell.Recently,single cell RNA sequencing has become a powerful tool to study cell dynamics and has significant progress in many areas.We propose a pipeline to cluster single cell and identify specific cell subtypes.Firstly,we present a novel method to d etermine the number of clusters of single cell RNA data.Next,we filter the data and do Linnorm normalization to make sure the data is on the same order of magnitude.Then we use TSNE to reduce the dimension and do the hierarchical clustering and the result can be evaluated by the ARI index.What's more,we utilize Kruskal-Wallis test and Wilcoxon-Mann-Whitney test to test the differential expressed gene.Finally,we use Fisher exactly test to determine the cell subtypes according to database.We set the database in the empirical analysis from the publicly data of six paper and we select two of them to show the result.we deal with the data and then analysis it according to the pipeline we proposed before.From the visualization of the cluster and the ARI index compared with other methods,it can be found that the number of clusters is very close to the actual results and the accuracy of the result is high.Then we determine the differential expressed gene with Wilcoxon rank test and determine the cell subt ypes with Fisher exactly test.In the end,we present some ideas for the advantages and disadvantages of our method and summarize the whole paper.