Abstract
Immunogenic antigen sets possess high potential for minimally invasive disease detection and monitoring. For various diseases, including cancer, appropriate antigen sets have already been detected in blood sera of patients. Typically, a large number of sera from diseased and unaffected persons is screened for the antigens of interest. Sophisticated statistical learning approaches are trained on the resulting data set to classify sera as either tumor or normal sera. We developed a web-based application, called ‘Seroreactivity Profile Classification Service’ (SePaCS) that enables clinical groups to carry out analyzes of training sets and predictions of unclassified seroreactivity profiles with minimal effort. SePaCS provides a broad range of classification methods: four versions of a Naïve Bayes Classifier, Support Vector Machines with a radial basis function kernel, Linear Discriminant Analysis, and Diagonal Discriminant Analysis. The computed results are summarized in a PDF file. We demonstrate the functionality of SePaCS exemplarily for meningioma, a generally benign intracranial tumor. As a second example, we evaluated SePaCS on glioma, a malignant brain tumor. SePaCS is freely available at http://www.bioinf.uni-sb.de/sepacs.