Stamoulis Catherine, Betensky Rebecca A, Mohapatra Gayatry, Louis David N
Department of Neurology, Harvard Medical School, Beth Israel Deaconess Medical Center, Boston, MA 02215, USA.
Annu Int Conf IEEE Eng Med Biol Soc. 2009;2009:6973-6. doi: 10.1109/IEMBS.2009.5333851.
We applied mode-decomposition and matched-filtering, both signal processing techniques used to increase the signal-to-noise ratio (SNR), to array CGH data of human meningioma DNA, in order to extract genomic regions of copy-number changes potentially associated with tumor progression. DNA segments from different chromosomes were decomposed into a small number of dominant components (modes), and low-amplitude modes were eliminated. The SNR of the entire segment was increased and it was possible to identify local changes in the data spatial structure, previously indistinguishable due to noise. We applied matched-filtering to the mode-reduced signals, using a normal DNA sequences (averaged over 50 healthy donors) as the template. The residual signals from this process were analyzed to identify disease-related copy number changes. We were able to identify distinct local changes at different chromosomes in patients with recurrent versus primary meningiomas.
我们将模式分解和匹配滤波这两种用于提高信噪比（SNR）的信号处理技术应用于人类脑膜瘤DNA的阵列比较基因组杂交（array CGH）数据，以提取可能与肿瘤进展相关的拷贝数变化的基因组区域。来自不同染色体的DNA片段被分解为少量的主导成分（模式），并消除了低振幅模式。整个片段的信噪比得到了提高，并且能够识别数据空间结构中的局部变化，这些变化之前因噪声而无法区分。我们使用正常DNA序列（对50名健康供体进行平均）作为模板，对模式简化后的信号应用匹配滤波。对该过程产生的残余信号进行分析，以识别与疾病相关的拷贝数变化。我们能够在复发性与原发性脑膜瘤患者的不同染色体上识别出明显的局部变化。
