IntroductionModern high-throughput genomic technologies have allowed the large-scale characterization of living organisms, involving the generation and interpretation of data at an unprecedented scale. Computational tools and mathematical algorithms have been created aiming to integrate, organize and mine the wealth of data generated. Technologies for the detection of different types of genomic alterations have been developed and applied to the analyses of living organisms and, in particular, cancer genomes. It is clear that studies based on a single technology are limited compared with the extent of knowledge that can be acquired using different technological platforms together. Hence, there is a need for systematic methodologies facilitating data management, visualization and integration. Such methodologies should aim to permit a proper analysis of the biological implications of findings, without sacrificing computational efficiency or mathematical and statistical rigour. With these purposes in mind, we have designed and implemented a data driven 3-state model for multidimensional data integration (3-MDI). ResultsLevel 1 and Level 2 data sets were pre-processed, and genes were selected in each platform based on the summary statistics presented in Table 1. Selected targets for mRNA expression and methylation were coded with our 3-state model {1,0, −1}, with 1 for upregulated or hypermethylation, 0 for no change, and −1 for downregulated or hypomethylation.The Level 3 somatic mutation data set was re-ordered by genes and coded in a 2-state format {0,1} rather than a {1,0, −1} 3-state format in order to avoid an ad hoc threshold for the classification of hypo/hyper mutation. The 2-state format represents the presence (1) or absence (0) of a mutation in a gene.Using these 3 platforms we could find up to 18 possible scenarios for further analysis (Fig. 1). Note that a single platform reports larger gene numbers as changing (e.g., pure gene expression cases {0,0, −1} with 1274 genes and {0,0,1} with 1418 genes), and numbers of genes changing are lower when combining multiple platforms and requiring congruent behavior, which is somehow exacerbated by to the fact that not all samples have measurements in all platforms.Background: Genomic technologies have allowed a large-scale molecular characterization of living organisms, involving the generation and interpretation of data at an unprecedented scale. Advanced platforms for the detection of different types of genomic alterations have been developed and applied to analyses of living organisms and, in particular, cancer genomes. It is clear now that studies based on a single platform are limited compared with the extent of knowledge gain possible when exploiting different platforms together. There is therefore a need for systematic methodologies facilitating data management, visualization, and integration.Materials and Methods: We present a 3-state model (3-MDI) that integrates several technological platforms, visualizing and prioritizing different biological scenarios, ...