表达谱芯片（Microarray）和转录组测序（RNA-seq）技术是基因表达检测的重要手段，但在实验设计、生物样品采集、RNA提取、文库构建方案、高通量测序平台、数据分析、实验室等方面容易引入误差，在数据中形成噪声和偏移，掩盖真实的生物学结果，影响结果可靠性，甚至导致错误的结论。申请人团队从2004年至今，使用相同的RNA标准样品，用不同的方法，在不同平台、实验室进行检查，已积累超过3千例表达谱数据，为研究系统噪声提供了独特的数据资源。申请人拟利用RNA标准样品，量化噪声来源，从差异基因检测和预测模型角度评估噪声特征和现有噪声消除方法对结果的影响，提出一套从实验设计到数据分析的有效控制噪声的指南，在此基础上建立参比数据库，为新的高通量技术和算法开发提供数据参考平台。本项目可提高结果的重复性和可靠性，对组学数据的有效应用和融合分析产生积极的影响，并最终为可靠的临床应用奠定基础。

Microarray and transcriptome sequencing (RNA-seq) technologies are powerful tools for measuring gene expression profiles. However, noises and biases are likely to be introduced into data due to inappropriate experimental design, biological sample acquisition, RNA extraction, library construction protocol, sequencing platform, data analysis pipeline, laboratory conditions, and so on. This becomes an important issue that may mask real biological signals of interest and may even lead to incorrect conclusions. Since 2004, we, by using the identical reference RNA samples, have accumulated large collections of expression data more than 3,000 runs, ranging from multiple protocols, platforms, and laboratory sites, providing a unique data resource for understanding the nature of measured noises. Using reference RNA samples, we propose to quantify the noises which have been introduced from design to data analysis, assess the effects of noises on the identification of differentially expressed genes and development of prediction model, evaluate the existing noise correction methods, put forward a set of guidelines from experimental design to data analysis for the purpose of effectively minimize noises and biases, and build a user-friendly database to benefit the new high-throughput techniques and algorithms. This project is expected to improve the repeatability and reliability of the genomic results, help data fusion, and build a solid foundation for proper application of gene expression profiles in precision medicine.