EFFECT OF MALDI-TOF SPECTRUM PREPROCESSING METHOD ON SERUM BIOMARKER DISCOVERY

这个子项目是许多研究子项目中的一个 由NIH/NCRR资助的中心赠款提供的资源。子项目和 研究者（PI）可能从另一个NIH来源获得了主要资金， 因此可以在其他CRISP条目中表示。所列机构为 研究中心，而研究中心不一定是研究者所在的机构。 我们收集了54例乳腺癌患者和443例多年无病患者的血清样本。在手术前和手术后7个月每隔一个月从癌症患者中收集样品。在不同时间点抽取对照组患者的血液。所有采血均在同一机构按照相同方案进行。从血清样品中提取白蛋白，并捕获附着于白蛋白的蛋白质。对这些提取的蛋白质进行高分辨率正交MALDI-TOF质谱。对所有样品进行技术重复，得到总共1750个光谱，每个光谱包含约500，000个数据点。来自所得光谱的峰用于构建样品的癌症与正常分类器。请参阅：利用指示翻译后修饰的峰对卵巢癌和CTCL血清样品进行分类的蛋白质组学模式。蛋白质组学2007年11月;7（22）：4045-52。光谱的预处理在正确识别样品之间的差异峰方面起着重要作用。目前还没有公认的标准光谱预处理方法。在这项研究中，我们建议将两种不同的预处理工作流程应用于整个数据集，并建立一个分类器来比较预处理方法对最终结果的影响。这两个预处理工作流程都是在Matlab中实现的，并可供公众使用。虽然Matlab可能不是最有效的实现，但在这种情况下，对方法的快速访问超过了与Matlab相关的小开销。我们将测试的两个工作流是：1。PrepMS，来自德克萨斯A&M统计部：http://www.stat.tamu.edu/~yuliya/prepMS.html 2。Matlab演示质谱预处理工作流程：http://www.mathworks.com/products/demos/shipping/bioinfo/mspreprodemo.html每个工作流程中的去噪、归一化和峰检测方法都不同。在成功完成这个小的比较研究后，我们将测试其他已发表的预处理方法。

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. We have collected serum samples from 54 patients diagnosed with breast cancer and 443 patients who have been disease free for several years. Samples were collected from cancer patients before surgery and at month intervals for seven months after surgery. Blood draws from control set of patients was drawn at different time points. All blood draws were performed in the same facility under the same protocol. Albumin was extracted from the serum samples and proteins attached to the albumin were captured. High resolution orthogonal MALDI-TOF mass spectrometry was performed on these extacted proteins. Technical repeats were performed for all samples yielding 1750 total spectra containing approximately 500,000 data points each. Peaks from the resulting spectra were used to build a cancer vs normal classifier for the samples. See: Proteomic patterns for classification of ovarian cancer and CTCL serum samples utilizing peak pairs indicative of post-translational modifications. Proteomics. 2007 Nov;7(22):4045-52. Preprocessing of the spectra plays an important role in correctly identifying differential peaks among samples. There is no currently accepted standard method of preprocessing spectra. In this study we propose to apply two different preprocessing workflows to the entire data set and build a classifier to compare the effect of the preprocessing method on final outcome. Both preprocessing workflows are implemented in Matlab and are available to the general public. Although Matlab may not be the most efficient implementation, ready access to the methods outweighs the small overhead associated with Matlab in this case. The two workflows we will test are: 1. PrepMS, from Texas A&M Department of Statistics: http://www.stat.tamu.edu/~yuliya/prepMS.html 2. Matlab demo mas spectrometry preprocessing workflow: http://www.mathworks.com/products/demos/shipping/bioinfo/mspreprodemo.html The method of denoising, normalization and peak detection differ in each workflow. Upon successful completion of this small comparison study we will test other published methods for preprocessing.