Salón de Grados E.T.S.I. Industriales- 18 abril, 12:45h.
Juan Luis Fernández Martínez
Group of Inverse Problems, Optimization and Machine Learning
Department of Applied Mathematics. University of Oviedo
Uncertainty Analysis of Phenotype Prediction Problems.
The Finisterrae project.
Genomic phenotype prediction is predicated on finding a set of genes that prospectively distinguishes a given phenotype. This kind of problem has a high underdetermined character since the number of monitored genetic probes markedly exceeds the number of collected samples (patients). This imbalance creates uncertainty as, potentially, there exist many equivalent genetic networks that could predict the phenotype with a similar accuracy, a feature that creates ambiguity in the characterization of the biological pathways since many genes are highly discriminatory. In top of that, a significant obstacle in the analysis of genetic data is the absence of a conceptual model that relates the different genes/probes to the class prediction between the set of genetic signatures and the set of classes in which the phenotype is divided.
In this talk I will show how we solve this interesting problems and some applications in cancer, rare and neurodegenerative diseases (FINISTERRAE PROJECT)