基于可搜索式单细胞拉曼数据库的单细胞高通量表型分析方法学研究

A Single-Cell Raman Spectrum (SCRS) harbors rich biochemical information about a cell and thus can be considered as an intrinsic, label-free “biochemical fingerprint” of a cell under a particular physiological state. By measuring and analyzing SCRS, Raman-activated Cell Sorting (RACS) is able to obtain key information about a cell such as its genetic background, physiological state and environmental conditions and then isolate the cell based on a target SCRS. However the lack of searchable SCRS reference database and the low throughput of current algorithms for SCRS comparison and search have hindered rapid expansion of RACS into various research and applied areas. Targeting these challenges, this project proposed to establish the prototype of a searchable SCRS database. Then a high-throughput and accurate SCRS database comparison, search and mining algorithm will be developed. A series of tests will be deployed to quantitatively investigate the influence of the various metadata on cell phenotype, and moreover to evaluate the accuracy and throughput of biologically annotating an unknown SCRS based on searching a reference SCRS database. Moreover, actual SCRS data from four model microbes with different size and shape will be used for performance evaluation and validation. Such a database-search based SCRS assessment model should provide a valuable method for storage and mining of single-cell Raman phenotype data and moreover should expedite the expansion of RACS into various application areas.

一个“单细胞拉曼图谱”（SCRS）蕴含着一个细胞丰富的生化信息，是细胞在特定生理状态下内在的、免外源标记的“生化指纹”。活体单细胞拉曼分选（RACS）通过测量SCRS及表型特征可获得细胞遗传背景、生理状态和环境条件等重要信息，并据此分离特定细胞。但可搜索式SCRS参照数据库的匮乏以及SCRS分析比对的低通量均限制了RACS的应用拓展。针对这些瓶颈，本项目计划设计一个可搜索式SCRS数据库，并开发一种大规模、高通量的微生物细胞表型（SCRS及与其对应的细胞形态等）数据的对比分析方法。将通过测试定量评价该数据库预测未知细胞之功能的准确性和通量，进而通过四种不同大小、形状的模式微生物的实际SCRS数据进行该计算流程的验证和评价。这一可搜索式单细胞拉曼数据库及其比对分析算法不仅为单细胞拉曼表型分析提供一种有效的信息存储和挖掘手段，而且将有力地推动RACS的拓展应用。

本项目基于微生物细胞表型（SCRS及与其对应的细胞形态等）数据，设计了微生物表型分析策略，并实现了完整的方法和软件开发，从而为单细胞拉曼表型分析提供一种有效的信息存储和挖掘手段，完成了本项目的研究计划。首先，本项目设计了可扩展、多层次的单细胞拉曼数据库和对应的数据预处理方法；其次，基于单细胞表型数据，开发了特征解析方法，降低特征空间维数，为后续表型分析提供具有代表性的数据支持；最后，开发了表型分析比对方法，建立微生物单细胞表型评价体系，并将本项目的相关研究成果在活体单细胞拉曼分选平台对微藻细胞进行实际应用。本项目共发表2篇学术论文，相关成果获得3项软件著作权授权，授权2项发明专利，申请2项发明专利，1项PCT专利。本项目不仅为细胞表型数据分析提供有力的支撑，用数据分析来推动RACS的发展，同时也为进一步充分有效地利用这些宝贵的大数据打下坚实的基础。