The aim of this research line is to better understand the dynamical behavior of biological systems. This research is on the interface of mathematics, computational science, and computational biology. The methods used are commonly in the form of sets of differential equations, agent-based models or combinations thereof. With these tools both the dynamical behavior of molecular networks and cellular interactions can be simulated. These simulations should have a predictive value and experimental data and experts’ knowledge should be incorporated when building the models. Examples of systems that are currently studied in this line are: optimal measurement protocols for enzyme kinetics (NWO-TA Moonlight) models of glucosinulate pathways in plants(NWO-LTP Plant-XR), agent based model for the germinal centre (in cooperation with AUMC group van Kampen) and  the interaction of the microbiome and the immune system in young infants (NWA-ORC  METAHEALTH). 

As a sidetrack we are also interested in developing interpretable neural nets by incorporation of knowledge and by constraining the number of connections in neural nets.  

Consortia involved

NWO-TA Moonlight, NWO-LTP Plant-XR and NWA-ORC METAHEALTH

With the increasing complexity of data generated in biological experiments it is becoming necessary to enhance the performance and applicability of existing statistical data analysis methods for improved resolution and interpretation of hidden biological information. For the integration of information from different genomics sources (e.g. mRNA, miRNA, proteomics or metabolomics) we develop low level data fusion methods that are able to distinguish the common information in the sets from the distinctive information that is specific for a given set of data. Furthermore we develop data analysis methods that are able to incorporate prior information. The incorporation of prior biological network knowledge to improve the analysis of complex transcriptomics and metabolomics data receives more and more attention in recent literature. A review on such methods was recently published by Reshetova et al in BMC Systems Biology, 8 Suppl 2. The figure below shows a slider approach where the prior information is incorporated into the model in a soft manner. The final model is based on a combination of the measured data and the prior knowledge.

In STATegra several low level data fusion methods are explored for their use in finding common and distinctive information from various data sources. As each method comes with its own assumptions and properties we explore their usefulness for given problems. Furthermore, we work on a common quality language for different omics data types using analytical figures of merit. Finally, we started to work on integration of different omics data sets using flux models that combine metabolic network information with measured metabolomics and genomics data.

Consortia involved

Personal Microbiome Health, NWO-ORC NGID

Microbial communities or “microbiomes” are ubiquitous. They interact naturally with complex organisms, such as humans, animals, and plants, with a big impact on their hosts and on the environment. However, our understanding of these interactions is still far from sufficient to predict, let alone steer, microbiomes to functionally enhance processes of interest, e.g., to improve human health or crop yield and resilience. One of the reasons is a lack of basic knowledge due to the challenges of observing microbes in their communities, and under natural or manipulated conditions. 

Meta-omics technologies aim to measure all representatives of a molecular class in a system that is comprised of several different species, e.g., all microbial genomes (the metagenome), or all microbial transcripts (the metatranscriptome). These technologies can therefore provide data on the identity and activity of microbial communities in situ.

This research line deals with the facilitation of biological interpretations through the application and development of bioinformatics methods that can extract meaningful information from large meta-omics data. This information can include occurrence or abundance of microbial taxa (e.g., from amplicon sequencing), the sequences and organization of microbial genomes (e.g., by metagenome binning), or activity measures of microbial genes. Another prerequisite for interpretation is the integration of data from different meta-omics levels. In the best case, borrowing information from one meta-omic level can improve the detection or interpretation on other meta-omics levels (multi-omics integration). Every meta-omics study creates new data. To make use of these efforts and to provide context, existing knowledge (e.g., in published literature) and databases (e.g., of published genomes) need to be accessible within bioinformatics workflows for the analysis and interpretation of new experiments and observations. 

Consortia involved 

Systems Biology Amsterdam , Personal Microbiome Health