It is rare that a single gene is sufficient to represent all aspects of genomic activity. Similarly, most common diseases cannot be explained by a mutations at a single locus. Since complex systems tend to be neither linear nor hierarchical in nature, but to have correlated components of unknown relative importance, the assumptions of traditional (parametric) multivariate statistical methods can rarely be justified on theoretical grounds. Empirical "validation" is not only problematic, but also time consuming. Here we demonstrates how bioinformatics tools, ranging from spreadsheets to grids, can enable u-statistics as a non-parametric alternative for scoring multivariate ordinal data. Applications are shown to improve assessment of genetic risk factors, quality control of microarrays and signal value estimation, scoring genomic profiles that best correlated with complex risk factors (cardiovascular diseases), and complex responses to an intervention (treatment of psoriasis).