gms | German Medical Science

54. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e.V. (GMDS)

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie

07. bis 10.09.2009, Essen

Statistical approaches for the inference of biological networks

Meeting Abstract

Search Medline for

  • Nicole Radde - Universität Stuttgart, Stuttgart

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie. 54. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie (gmds). Essen, 07.-10.09.2009. Düsseldorf: German Medical Science GMS Publishing House; 2009. Doc09gmds113

DOI: 10.3205/09gmds113, URN: urn:nbn:de:0183-09gmds1135

Published: September 2, 2009

© 2009 Radde.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc-nd/3.0/deed.en). You are free: to Share – to copy, distribute and transmit the work, provided the original author and source are credited.


Outline

Text

The problem of reverse engineering intracellular networks from experimental data often results in a highly nonlinear and underdetermined optimization problem. Data are usually noisy and sparse, and the systems are intrinsically stochastic. Therefore, a stochastic modeling framework and statistical approaches for network inference are appropriate in this setting, since they naturally take uncertainties and measurement errors into account.

In my talk I compare likelihood functions of different time-discrete stochastic models which have been suggested to capture stochastic effects in biological network models. I propose to classify those models into three groups, according to the interpretation of the origin of stochasticity. General expressions for likelihoods are developed, and a comparison of those across the groups is provided. This method also suggests a way to separate noise in biological systems, which is illustrated on a small sample network. Here, a challenge is the investigation of marginal likelihoods, which is computationally expensive and requires efficient sampling methods.