Systems biology focus on system level analysis of molecular biology by modeling biological properties from gene function to organism phe- notypes as emergent properties arising from interacting genes, proteins and metabolites. We apply systems biology approach to reverse-en- gineer the regulatory networks of plants from transcriptomics, proteomics and metabolomics (omics) data. We then use these networks as the basis for investigating the complexity and evolution of gene regulation across diverse plant species and for predicting the effects of artificial perturbation in transgenic plants. Our network models can be used to computationally generate testable hypothesis and have been applied, for example, to select candidate genes for perturbation experiments.
The recent revolution in omics technology has enabled re- searchers to move beyond dissecting biological systems one gene at a time, and instead modeling interactions of multiple genes, proteins and metabolites required to understand complex biological properties.The focus on interactions is the hallmark of systems biology, and requires integrating massive amount of heterogeneous data sources to curb the combinatorial explosion resulting from studying more than one gene at a time.To infer network models from data, we use a technique from computer science called machine learning.Machine learning infers general models form characterized observations (examples) and can be used both to explain underlying patterns in data and to provide predictions for new, uncharacterized observations.
The focus on gene networks rather than on individual genes has, for example, allowed us to study the complexity of gene regulation in aspen leaves and wood.We found that a number of relevant regulators in these systems could only be identified when considering interactions of regulators such as AND logics. We have also compared networks across plant species and shown, for example, that gene centrality in regulatory networks tends to be conserved. Moreover, by studying the conservation of gene neighborhoods across species we can more confidently identify the most likely functional orthologs among several predicted candidates. Finally, we have demonstrated the power of using network for predicting the molecular effects of perturbation experiments and for explaining the observed phenotypes in transgenic trees.
We are developing a number of online tools for facilitating systems biology analysis in plants. ComPlEx is a portal for Com- parative analysis of Plant Expression networks, and PopGenIE/ ConGenIE (Populus/Conifer Genome Integrative Explorer) now include network tools for performing co-expression anal- ysis. All tools are available from the PlantGenIE web resource (http://plantgenie.org).
- Netotea S, Sundell D, Street NR and Hvidsten TR (2014). ComPlEx: Conservation and divergence of co-expression networks in A. thaliana, Populus and O. sativa. BMC Genomics 15:106.
- Street N, Jansson S and Hvidsten TR (2011). A systems biology model of the regulatory network in Populus leaves reveals interacting regulators and conserved regulation. BMC Plant Biology 11: 13.
- Önskog J, Freyhult E, Landfors M, Rydén P and Hvidsten TR (2011). Classification of microarrays; synergistic effects between normalization, gene selection and machine learning. BMC Bioinformatics 12: 390.
- Björkholm P, Daniluk P, Kryshtafovych A, Fidelis K, Andersson R and Hvidsten TR (2009). Using multi-data hidden Markov models trained on local neighborhoods of protein structure to predict residue-residue contacts. Bioinformatics 25: 1264-1270.
- Hvidsten TR, Wilczynski B, Kryshtafovych A, Tiuryn J, Komorowski J and Fidelis K (2005). Discovering regulatory binding site modules using rule-based learning. Genome Research 15: 856-66.