You are on page 1of 2

Prof.

Dagmar Iber, D-BSSE

Mathematical Modeling in Systems Biology

AS2011

Exercise 8
1. Scale-free networks (40%). The Barabasi-Albert model is an algorithm that can be used to generate scale-free networks. This method builds up the full network sequentially starting from a seed initial network. At every step one additional node is added, and the network size increases by one. Each new node is automatically linked to the network with one edge based on a preferential attachment mechanism. Therefore the probability Pi of every new node to be linked to one of the already existing nodes i is: Di N j =1 Dj

Pi =

(1)

where Di corresponds to the degree of the node i and N corresponds to the number of nodes of the network. Obviously, the degree of each node of the seed network must be at least one, otherwise it will remain disconnected throughout this process. a) Use a seed network of size 5 and implement this algorithm in matlab to create a scale-free network of size 100. Plot your result. b) Generate a second network with the same total number of nodes and edges, without the preferential attachment this time. To do this, just assign sequentially random edges until the two networks have the same total number of edges. Plot this graph and comment on the dierence with the scale-free graph. c) Plot the corresponding degree distributions as histograms. In which case do you observe hubs? Comment on their biological role. 2. Rule-based modeling (60%). Use the rule-based software BioNetGen to implement the following systems: i) Protein A has three binding sites x, y, z . Proteins B ,C and D can interact (bind/unbind) with protein A at the x,y and z binding sites respectively. Fill in the missing parts of the given .bngl le to generate the full system and plot (for tspan = 0:100) all the states, as well as the following observables : total A, free A, A that is at least bound with B , Free B , bound B , bound C , free D. ii) Consider a bivalent ligand L(l, l) and a bivalent receptor R(r, r). Sequential association leads to the formation of large complexes. We neglect the formation of rings. As long as the receptor has at least one free binding site R(r), it can interact (bind/unbind) with a ligand that has exactly two free binding sites L(l, l), or just one free binding site L(l, l!+). Create the .bngl le similarly to (i) just with the two rules described above. In order to set an upper limit for your system size, you can require that you have complexes with up to six receptor, or six ligands. To do this you have to modify the actions: actions generate network({overwrite=>1,max stoich=>{R=>6,L=>6}}) writeMfile({}); 1

As observables dene the free ligand, the free receptor, as well as the receptor-dimers, receptor-trimers, and complexes containing more than four receptors. Use order indicated below: begin observables ...freeL... ...freeR... RDimers R==2 RTrimers R==3 RTetramers R==4 gt4mers R>4 end observables Create a new runle runfile ii.m based on the runfile i.m, in order to integrate your system (tspan = 0:10) and plot all the states and observables. As initial concentrations use: R0 = 300 for the receptor, and L0 = 100 for the ligand. For the binding rates use kp1 = 0.001, kp2 = 0.002 (positive cooperative binding), and for the unbinding rates km1 = km2 = 0.01 Guidelines: (1.) Fill in the missing parts in ex8 1.m (2.) Unzip the compressed le and you will see that you have a folder with the version 2.17 of BioNetGen (BioNetGen217) as well as a template .bngl le for (i). After you ll in the missing parts (...) you can compile it from the command line (terminal). So while being on the directory of the unzipped le you would type: ./BioNetGen217/Perl2/BNG2.pl filename.bngl This will generate a .net and a .m le. convertBNGL.m will use these les to create your model structure (as indicated in the included runle for (i) : runfile i.m ). For (ii) you have to create the .bngl le and create a new runle runfile ii.m that is given for (i) so that it matches your new model. The calcObs i.m, calcObs ii.m les (given) are required for the calculation of your observables, and they are specic for each one of the two cases.

You might also like