Honors Thesis Presentation
Student: John Hocter
Title: Integrative Genomic Analysis for Improved Prediction
Advances in high-throughput technologies have led to the acquisition of various types of -omic data on the same biological samples. Each data type provides a snapshot of the molecular processes involved in a particular phenotype. An integrative -omic analysis, by combining the complementary information from the different data types, can provide a better understanding of the complex biological mechanisms involved in the etiology or progression of a disease.
In this thesis, I focused on ovarian cancer and used data from The Cancer Genome Atlas (TCGA) project to build predictive models of survival. I investigated various models ranging from using clinical data alone to an integrative model combining clinical and various genomic data (gene expression, methylation, copy number variation). I used variable selection in Cox survival models, and in particular the lasso technique, to identify relevant biomarkers. To evaluate the predictive performance of the models, I used time-dependent ROC curves and concordance index measures based on repeated k-fold cross-validation. I demonstrate that there is a substantial gain in prediction when integrating the various genomic data sets to the clinical data.
Monday, April 24, 2017 at 11:00am to 11:30am
St. Mary's Hall, 326
3700 Reservoir Road, N.W., Washington