Poster #75 - Jason Liu

A Novel Algorithm for Active Module Aggregation Based on the Earth Mover's Distance

Liu, J.1, Xu, M. PhD2, Xing J., PhD1,1 - Department of Genetics, Rutgers, The State University of New Jersey, Piscataway, NJ, USA., 2 - Department of Statistics, Rutgers, The State University of New Jersey, Piscataway, NJ, USA

Gene-gene interaction networks play a central role in -omics analysis. These networks can be combined with transcriptomic data to better understand how candidate genes interact with one another, leading to their prioritization. However, biological networks can become too large to manually analyze, motivating effective ways to reduce their complexity. One approach to this is Active Module Identification (AMI), which seeks to identify regions of interest within the network based on connectivity and gene expression. Many algorithms have been developed to this end, and previous studies suggest that AMI algorithms identify biologically distinct valid signals on the same dataset. Thus, effective ways to aggregate the outputs of these algorithms are needed. In this study, we propose a module aggregation method that optimizes an objective based on the Earth Mover's Distance, which measures the biological similarity between modules. The output is a set of disjoint modules that can be easily interpreted, and represents the overall agreement between the input modules. Our method will allow researchers to more completely identify the biological signal in their data.