Scalable dynamic fuzzy biomolecular network models for large scale biology

Abstract

Fuzzy logic is an effective language for models that interpret large scale, high throughput molecular biology experiments, including genomics, proteomics,metabolomics, and inhibitor screening. Two important principles apply for biological system modeling: (1) In the post-genome era, the development of novel molecular diagnostics and therapeutics requires interpreting the complex results of high-throughput multiplexed experiments, and a framework to efficiently and rapidly design hypothesis-driven experiments. (2) Biomolecular data are typically noisy and semi-quantitative, in particular because of the typical fluorescence output of high throughput experiments. Fuzzy biomolecular network models coupled with hypothesis generation strategies address these needs. In this chapter, we describe an integrated, data-driven method for extracting system models from data and generating hypotheses for experimental design. The method is based on scalable, linear relationships between nodes of a biomolecular network, representing the expression of genes, proteins, and/or metabolites. Data from high-throughput are fuzzified using a universal normalization method. Best-fitting models are generated through an evolutionary algorithm, and disagreements between plausible hypothetical network models are used as the basis for identifying experimental designs. The result is a modeling and simulation framework that can be easily integrated with text-based and graphical biological knowledge contained within existing literature and databases. © 2009 Springer-Verlag Berlin Heidelberg.