SVD Based Feature Selection and Sample Classification of Proteomic Data

Feature selection becomes a central task when 'signature' profiles specific to a pathological status have to be extracted from high dimensional gene expression or proteomic data. In the present paper, we propose a feature selection method based on Singular Value Decomposition (SVD) and apply it to SELDI-TOF/MS proteomic data from a cohort of Type 2 Diabetics affected by Glomerulosclerosis and Membranous Nephropathy. We have selected a profile composed of 24 proteins that seems to be an effective signature for the pathology at hand, allowing to efficiently discriminate between the considered subtype of diabetes.