Rogers Wade T, Holyst Herbert A
Department of Pathology and Laboratory Medicine, School of Medicine, University of Pennsylvania, 207 John Morgan Bldg., Philadelphia, PA 19104-6082, USA.
Adv Bioinformatics. 2009;2009:193947. doi: 10.1155/2009/193947. Epub 2009 Sep 24.
A new software package called flowFP for the analysis of flow cytometry data is introduced. The package, which is tightly integrated with other Bioconductor software for analysis of flow cytometry, provides tools to transform raw flow cytometry data into a form suitable for direct input into conventional statistical analysis and empirical modeling software tools. The approach of flowFP is to generate a description of the multivariate probability distribution function of flow cytometry data in the form of a "fingerprint." As such, it is independent of a presumptive functional form for the distribution, in contrast with model-based methods such as Gaussian Mixture Modeling. FlowFP is computationally efficient and able to handle extremely large flow cytometry data sets of arbitrary dimensionality. Algorithms and software implementation of the package are described. Use of the software is exemplified with applications to data quality control and to the automated classification of Acute Myeloid Leukemia.
介绍了一种名为flowFP的用于分析流式细胞术数据的新软件包。该软件包与其他用于流式细胞术分析的Bioconductor软件紧密集成，提供了将原始流式细胞术数据转换为适合直接输入到传统统计分析和经验建模软件工具的形式的工具。flowFP的方法是以“指纹”的形式生成流式细胞术数据多元概率分布函数的描述。因此，与基于模型的方法（如高斯混合建模）不同，它独立于分布的假定函数形式。FlowFP计算效率高，能够处理任意维度的极大流式细胞术数据集。描述了该软件包的算法和软件实现。通过数据质量控制和急性髓细胞白血病自动分类的应用举例说明了该软件的使用。
