Compensation Automation for HIV Vaccine Development

描述(由申请人提供)：许多基础和临床研究，特别是在艾滋病毒领域，依赖于流式细胞术来收集和分析所需的研究数据。在流式细胞仪仪器和试剂可用性方面的最新进展为处理大量样本和利用该技术在单细胞水平上解决更广泛的问题开辟了道路，例如，在艾滋病毒感染者和非感染者的刺激和非刺激外周血淋巴细胞样本中，检测蛋白质磷酸化、细胞因子、趋化因子和酶的产生，这是一种在“精细”亚群水平上衡量T细胞激活的指标。然而，我们发明/开发了当今流式细胞仪中包含的许多基本技术，并运营了一个主要的基础和临床研究实验室多年，还领导了一个主要的基础和临床研究实验室和斯坦福大学的主要流式细胞仪服务中心，我们敏锐地意识到，新的FACS仪器(和旧的)的能力远远超过大多数使用这些仪器的实验室目前的数据处理能力。这是艾滋病毒研究中的一个特殊问题，也是我们实验室特别感兴趣的一个领域，我们对处理流动数据所涉及的困难有直接的经验。认识到这一问题，我们建议在这里开发软件，使流式细胞仪数据的标准化和荧光补偿完全自动化。这些初始计算步骤准备用于分析的原始流动数据，非常耗时且需要大量技能。然而，我们的初步研究表明，它们可以由软件可靠且一致地执行，并且在更注重数据质量的情况下，由完成这些操作的软件而无需用户干预。我们建议开发这一软件，一旦开发完成，将通过合作的商业来源，迅速提供给艾滋病毒研究界和其他方面的“免费软件”。该项目将包括开发和验证一种基于综合数据模型评估荧光补偿的新方法。除了完全自动化之外，这种方法比目前提供误差估计和其他质量保证信息以确认自动化分析结果是可靠的方法更有优势。公共卫生相关性：我们提出的研究重点是开发对流式细胞仪数据进行预处理的方法。为了成功，这些工具必须在没有调查人员干预的情况下执行三个基本功能：1)它们必须将数据“标准化”，以便在不同日期或不同地点收集数据时可以进行比较；2)它们必须对数据进行“补偿”，以纠正每个荧光通道中获得的值与其他通道中读取的染料的光谱重叠；以及3)因为数据质量是核心，它们必须评估原始数据和处理后的数据，并在数据质量受到影响时向用户发出警告。最后，当分析涉及区分暗染细胞和未染细胞时，这些工具应该能够计算虚拟FMO分布，从而允许使用染色样本的数据来确定进行这种区分的适当阈值。基本上，在数据准备阶段结束时，数据应该准备好进行分析，调查人员应该确信处理后的数据是可靠的，或者知道为什么不是！

DESCRIPTION (provided by applicant): Many basic and clinical studies, particularly in the HIV arena, rely on flow cytometry for collecting and analysizing the study data needed. Recent advances in flow cytometry instrumentation and reagent availability have opened the way to processing larger numbers of samples and to using the technology to address broader questions at the single cell level, e.g., examination of protein phosphorylation, cytokine, chemokine and enzyme production, a measure of T cell activation at the "fine" subset level, in stimulated and non-stimulated peripheral blood lymphocyte samples from HIV-infected and non-infected subjects. However, having invented/developed much of the basic technology incorporated in today's flow cytometry instruments, and also having run a major basic and clinical research laboratory for many years, and also having led a major basic and clinical research laboratory and the main flow cytometry service center at Stanford, we are keenly aware keenly aware that the capabilities of the new FACS instruments (and the older ones) far outstrip the current data handling capabilites of most laboratories employing these instruments. This is a particular problem in HIV research, an area of special interest in our laboratory and one in which we have direct experience with the difficulties involved in processing flow data. Recognizing this problem, we propose here to develop software that will fully automate the standardization and fluorescence compensation of flow cytometry data. These initial computation steps, which prepare raw flow data for analysis, are very time consuming and require substantial skill However, our preliminary studies indicate that they can be executed reliably and consistently by software, and with greater attention to data quality, by software that completes these operations without user intervention. We propose to develop this software and, once developed, to make it rapidly available to the HIV research community and beyond as "freeware" and through cooperating commercial sources. This project will involve development and verification of a new method for evaluating fluorescence compensation based on a comprehensive data model. In addition to full automation this method will have the advantage over the current methods of providing error estimates and other quality assurance information to confirm that the results of the automated analysis are reliable. PUBLIC HEALTH RELEVANCE: The studies we propose focus on development of methods for pre-processing flow cytometry data. To be successful, these tools must perform three basic functions without investigator intervention: 1) they must "standardize" the data so that it can be compared when collected on different days or at different sites; 2) they must "compensate" the data to correct the values obtained in each fluorescence channel for spectral overlap from dyes read in other channels; and, 3) because data quality is central, they must evaluate the raw and processed data and warn the user if data quality has been compromised. Finally, when analyses involve distinguishing between dully stained and unstained cells, the tools should be able to compute virtual FMO distributions that allow data for a stained sample to be used to determine appropriate thresholds for making this distinction. Basically, at the close of the data preparation phase, the data should be ready for analysis and the investigator should either be confident that the processed data is reliable or know why it is not!