Discipline: Technology and Engineering
Subcategory: Cancer Research
Camille Marrero Laboy - University of Puerto Rico Mayagüez Campus
Co-Author(s): Mauricio Cabrera and Clara Isaza, University of Puerto Rico Mayagüez Campus
Breast cancer (BC) is the most commonly diagnosed cancer among women in the US. Abnormal patterns of genetic expression are among the multiple factors that contribute to the occurrence of BC. This work aims to provide evidence of genes that are deemed important to the presence and development of this illness, as well as their coordinated behavior. These genes were determined by the application of Multiple Criteria Optimization (MCO). The novelty in this study was the characterization of the signaling path through the Traveling Salesman Problem (TSP) combinatorial optimization formulation. This analysis was done with the use of a microarray database related to BC by finding coordinated behavior among the expression levels of different genes in various types of breast cancer metastasis. To carry out the database analysis we applied the MCO; this is an optimization problem that finds a set of solutions corresponding to the best possible relation among two or more conflicting performance measures under study. The genetic expression in each type of BC metastasis was quantified using the metrics of difference of means and difference of medians. For the analysis, the MCO solutions were the genes that had associated the greatest differences in the selected metrics. Potential biomarkers were selected with an intragroup comparison. By finding the genes with the most significant change in expression in each individual metastasis category we were able to compare them with the other potential biomarkers of the other types of metastatic BC.The resulting structure is, indeed, a global optimum and, thus, should provide a very competitive manner to direct biological search and organize the evidence of the roles and dynamics of genes in BC. The database with GEO identifier GDS4761 was used for this analysis. The database contains global gene expression for 51,562 genes from 91 samples, however only 88 samples were considered as significant data. The cancerous samples were divided as follows: 5 bone metastasis, 11 local in the breast metastasis, 16 liver metastasis, 39 lymph node metastasis and 17 skin metastasis. Potential biomarkers obtained with the MCO demonstrated patterns of expression in 10 comparative pairs of metastases generated with the previously mentioned metastases categories. This study aims to detect important genes in breast cancer, and offer a structure of maximal correlation among them to guide the search of biological evidence for the role and dynamics of these genes. In future studies a TSP approach will be applied where linear correlation values between the expression of different genes can be used to construct a signaling pathway. Uniprot Database and the Database for Annotation, Visualization, and Integrated Discovery (DAVID) will be used to further understand the effect and function of the patterns observed in the preliminary results.
Funder Acknowledgement(s): C. Marrero gratefully acknowledges the support of PRLSAMP for the continuation of this work.
Faculty Advisor: Mauricio Cabrera, mauricio.cabrera@gmail.com
Role: My responsibilities included selection of the database, application of the method, and analysis of the results.