Additive ODE Models for Reconstructing Biological Networks

• Activity Number: 569
• Type: Contributed
• Date/Time: Wednesday, August 6, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): James Henderson*+ and George Michailidis
• Companies: University of Michigan and University of Michigan
• Keywords: network reconstruction ; dynamic systems ; ODEs ; time series

Abstract

Network representations of biological systems are widespread and reconstructing unknown networks from data is a focal problem in computational biology. For example, regulatory relationships among genes are commonly represented as directed networks with edges connecting influential genes to their targets. Reconstructing such networks from data is a challenging problem receiving much attention. There is a particular need for approaches tailored to time-series data and not reliant on direct intervention experiments. We introduce a new approach to reconstructing directed networks using dynamic systems models. Our approach generalizes commonly used linear and nonlinear ODE models by extending the functional class from parametric to nonparametric models. Concomitantly we limit the complexity by imposing an additive structure on the nonparametric slope functions, which are also required to be smooth. We show the utility of the method by reconstructing networks using simulated data from a computational model for a gene network regulating the pluripotency of mouse embryonic stem cells.