So, improvement of existing methods or development of new methods is needed for the analysis of gene expression microarray data. Many gene expression signatures have been identified in recent years for accurate classification of tumor

subtypes [16–19]. It has been indicated that rational use of the available bioinformation can not only effectively remove or suppress noise in gene chips, but also avoid one-sided results of separate experiment. However, a relatively few attempts have been aware of the importance of prior information in cancer classification [20–22]. Lung cancer is one of the leading causes of cancer death worldwide [23–26], can be classified broadly into small cell lung buy 5-Fluoracil cancer (SCLC) and non-small cell lung cancer (NSCLC), and adenocarcinoma

is the most common form of lung cancer. Because in China the cigarette smoking rate continues to be at a high level [27], a peak in lung cancer incidence is still expected [28]. Therefore, only lung cancer gene expression microarray dataset was selected in the present study. In summary, together with the application of support vector machine as the discriminant approach and PAM as the feature gene selection method, we selleck propose one method that incorporates prior knowledge into cancer classification based on gene expression data. Our goal is to improve classification accuracy

based on the publicly available lung cancer microarray dataset [29]. Methods Microarray dataset In the present study, we analyzed Ribonucleotide reductase the well-known and publicly available microarray dataset, malignant pleural mesothelioma and lung adenocarcinoma gene expression database http://​www.​chestsurg.​org/​publications/​2002-microarray.​aspx[29]. This Affymetrix Human GeneAtlas U95Av2 microarray dataset contains 12 533 genes’ expression profiles of 31 malignant pleural mesothelioma (MPM) and 150 lung adenocarcinomas (ADCA, published in a previous study [30]), aims to test expression ratio-based analysis to differentiating between MPM and lung cancer. In this dataset, a training set consisted of 16 ADCA and 16 MPM samples. Microarray data preprocessing The absolute values of the raw data were used, then they were normalized by natural logarithm transformation. This preprocessing procedure was performed by using R statistical software version 2.80 (R foundation for Statistical Computer, Vienna, Austria). Gene selection via PAM Prediction analysis for microarrays (PAM, also known as Nearest Shrunken Centroids) is a clustering technique used for classification, it uses gene expression data to calculate the shrunken centroid for each class and then predicts which class an unknown sample would fall into based on the nearest shrunken centroid.