Multiple Criteria Analysis of High-Throughput Biological Experiments

On previous works, our research group proposed that the detection of potential biomarkers from the analysis of microarray experiments were modeled as a multiple criteria optimization problem. The benefits of this representation are the preservation of objectivity in the analysis and the possibility of analyzing experiments with incommensurable units simultaneously.  The technique proposed to solve the resulting optimization problem was Data Envelopment Analysis (DEA). This work proposes the adoption of a technique based on the concept of Pareto-dominance (PD). PD has a better resolution than DEA, although at the cost of being more computationally intensive. The aims are to, first, implement PD to compare its results vs. those of DEA in different cancer scenarios. Then, a scheme to make the PD more computationally tractable will be designed through the use of clustering techniques. If successful, this project will enhance the multiple criteria analysis capabilities of microarray experiments and high-throughput biological experiments with a similar matrix structure, such as reverse-phase protein arrays.