Design Of Web-Based Platform For Single-Cell Transcriptome Data Analysis

With the development of scRNA-seq technology,research of gene expression has reached an unprecedented level.Lot of scRNA-seq data analysis tools have been developed which are built on different programming languages,making it difficult for bioinformaticians to keep up with the increasingly complex analytical processes and rapidly evolving methods.This has called new demands for standardization,big data processing,personalized analysis,process simplification,and so on.Though many data analysis platforms have been developed,still many problems exist.On the one hand,some platforms are built on inappropriate frameworks which reduce their extendibility and portability.When deploying the platform on a new computer,rebuilding of the platform system from source codes is typically required.On the other hand,most platforms lack of an upstream data analysis module or require overly complex operation processes to complete downstream data analysis with patchy visualization results.It thus becomes important and urgent to remove the barrier for biologists who have insufficient expertise in computer technologies,and to integrate multiple analysis workflows properly to reduce the workload for configuring the computing environment and reduce the learning cost.Based on the above considerations,this thesis presents a design to build a platform which can analyze scRNA-seq data from upstream to downstream with appropriate visualization and pleasant user experience.From a user’s point of view,the presented platform realizes "one station" upstream analysis using the Nextflow framework.For downstream analysis,it integrates cell annotation,PCA,t-SNE/UMAP,DE algorithm based on a hurdle model,the KNN+Louvain clustering algorithm,to construct a step-bystep and repeatable experiment process.This makes the platform simple but comprehensive,user friendly and visualized.Its usability enables biologists to focus on the biological research behind data without complicated installation and configuration procedures.In terms of data analysis and web application,the presented platform has balanced diversity and universality of the tools deployed.The backend and the frontend are separated based on MVVM,empowering this platform with good scalability to adapt to the fast evolving for analysis algorithms and tools,and can be easily deployed and transported with new computer hardwares.