This Project is linked to the course MSB1014 Network Biology in the Master's Systems Biology & Bioinformatics at Maastricht University.
To run the network creation script the following Human1 model files (release version 1.19.0) are needed: metabolites, reactions and the model as a .txt itself. The .sif version of the model created in cytoscape is already provided.
To create the HumanGEM sif file the xml/sbml version of the Human1 [1] model was loaded into cytoscape [2] using the cy3sbml plugin [3]. All nodes except species and reactions were removed and the nodes were renamed according to the schema "type:id". Afterwards the model was exported as a .sif file. The network was created in R using igraph (see network creation) and further annotated using the Human1 model files.
Genome-scale metabolic networks capture the metabolism of a specific organism in a single directed network. One challenge when analyzing these networks is what is referred to as “currency cofactors”, which include ATP/ADP, NAD(H) and many more. These cofactors are involved in a lot of reactions due to their fundamental functionality and create shortcuts between different parts of the network that are not necessarily functionally related. This distortion of clusters makes it hard to detect biologically meaningful modules, so any clustering approach has to take them into account [4]. This Project will take the state of the art human metabolic network Human1 [1] and explore how different approaches to limit currency cofactor influence alters a WalkTrap [5] clustering approach in finding biologically meaningful modules.
Determine how handling currency cofactors affect cluster detection and the biological coherence of these clusters in the Human1 metabolic network.
Research questions:
- Will removing or changing the edge-weight of currency cofactors affect clustering more?
- Are clusters more biologically coherent after dealing with currency cofactors?
The Human1 metabolic model will be used as a baseline for three different networks variants: Unaltered, currency cofactors removed, currency cofactors down weighted. The set of currency cofactors are selected based on the definition of the cy3sbml plugin [3]. Down-weighting of edges will be scaled by frequency of the cofactor in the network and degree. WalkTrap [5] will be used as a simple clustering algorithm on the three networks to find communities. Afterwards, pathway enrichment analysis is employed to assess the influence of handling “currency cofactors” and estimate the biological coherence of the found communities. The focus of the enrichment analysis is only on the 10 largest communities that are found. The analysis will be done in R.
1. Robinson JL, Kocabaş P, Wang H, Cholley PE, Cook D, Nilsson A, et al. An Atlas of Human Metabolism. Sci Signal. 2020 Mar 24;13(624):eaaz1482.
2. Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003 Nov;13(11):2498–504.
3. König M, Dräger A, Holzhütter HG. CySBML: a Cytoscape plugin for SBML. Bioinforma Oxf Engl. 2012 Sept 15;28(18):2402–3.
4. Huss M, Holme P. Currency and commodity metabolites: their identification and relation to the modularity of metabolic networks. IET Syst Biol. 2007 Sept 17;1(5):280–5.
5. Petrochilos D, Shojaie A, Gennari J, Abernethy N. Using random walks to identify cancer-associated modules in expression data. BioData Min. 2013 Oct 15;6(1):17.