The invention provides with means to predict, in subjects affected by type 2 diabetes (T2D), the probability of developing complications related to the disease. The invention involves 1) use of genetic features (SNPs, STRs, or other genomic markers) together with other chromosomal features and phenotypic information to establish a patient profile specifically developed for prediction of complications of T2D 2) use of a set of SNPs allowing to discriminate between individuals according to their descent. The invention also provides with methods for characterizing and selecting, within a population of subjects with type-2 diabetes, subjects that are suited for clinical trials based on the identification of one or more genetic features. Also described are combinations and kits for carrying out the above-described methods.