Application of Multivariate Data Analysis for the Classification of Two Dimensional Gel Images in Neuroproteomics

Two-dimensional gel electrophoresis (2DE) still plays a key role in proteomics for exploring the protein content of complex biological mixtures. However, the development of fully automatic strategies in extracting interpretable information from gel images is still a challenging task. In this work, we present a computational strategy aiming at an automatic classification of the discriminant patterns emerging from separation images intended as fingerprints of the correspondent biological conditions. The method was applied to gel images acquired in a study on motor neuron diseases: 33 2DE maps generated from samples of cerebrospinal fluid were processed (26 pathologic and 7 control subjects). Quantitative image descriptors were extracted and fitted to a partial least squares-discriminant analysis (PLSDA) assessing the chance to classify the samples. Moreover, the model was able to identify gel areas that most differ through the clinical categories. Combining multivariate statistical techniques with 2DE may represent a valid tool to extract informative protein patterns. This kind of approach can contribute to the development of a system of screening to discriminate different clinical conditions on the basis of the overall patterns emerging from the maps, representing a useful complementary analysis in the routine of a proteomic laboratory.