Methods for data integration in complex human diseases
Wednesday, November 18, 2015
3:30 pm - 4:30 pm
Gross Hall 330
Anna Goldenberg-University of Toronto
12noon-postdoc/grad student lunch 3:30pm seminar How can we combine multiple types of measurements to create a comprehensive view of a given disease? In this talk I will introduce two approaches that address complementary aspects of this question. First, I will present our recently developed Similarity Network Fusion (SNF) method to integrate genomic and other types of data for the same set of patients. Combining three biological data types, SNF substantially outperformed single data type analysis and other integrative approaches to identify cancer subtypes in five cancers. Patient networks made us think deeper about the problem of feature (e.g. gene) selection, diagnosis and prognosis for each individual patient. I will thus present an alternative test for differential gene analysis and a novel patient-network based regularization of the Cox survival model. In the second part of the talk, I will present a model that aims to recover disease mechanisms. To increase the power of identifying genes associated with diseases and to account for other potential sources of protein function aberrations, we propose a novel factor-graph based model, where much of the biological knowledge is incorporated through factors and priors. Our extensive simulations show that our method has superior sensitivity and precision compared to variant-aggregating and differential expression methods. Our integrative approach was able to identify important genes in breast cancer.