A very fast and accurate method for calling aberrations in array-CGH data
A very fast and accurate method for calling aberrations in array-CGH data
复制标题
DOI：
10.1093/biostatistics/kxq008
复制
发表时间：
2010-07-01
期刊：
影响因子：
2.1
通讯作者：
Magi, Alberto
中科院分区：
文献类型：
--
作者：
Benelli, Matteo;Marseglia, Giuseppina;Magi, Alberto
阵列比较基因组杂交(ACGH)是一种可以检测和定位基因组变化的微阵列技术。ACGH数据分析的标准工作流程包括两个步骤：通过分割算法(断点识别)检测拷贝数改变的区域的边界，然后使用概率框架(调用过程)将每个区域标记为丢失、中性或收益。本文介绍了一种新的基于截断正态分布的调用过程，称为FastCall，其目的是以非常快速和准确的方式给出分割的aCGH数据的像差概率。无论是在人工合成的aCGH数据上，还是在真实的aCGH数据上，FastCall在分类精度和运行时间方面都取得了优异的性能。
Array comparative genomic hybridization (aCGH) is a microarray technology that allows one to detect and map genomic alterations. The standard workflow of the aCGH data analysis consists of 2 steps: detecting the boundaries of the regions of changed copy number by means of a segmentation algorithm (break point identification) and then labeling each region as loss, neutral, or gain with a probabilistic framework (calling procedure). In this paper, we introduce a novel calling procedure based on a mixture of truncated normal distributions, named FastCall, that aims to give aberration probabilities to segmented aCGH data in a very fast and accurate way. Both on synthetic and real aCGH data, FastCall obtains excellent performances in terms of classification accuracy and running time.