BioNetGen to SSC translator
From BioNetWiki
Contents |
About
Stochastic Simulation Compiler, SSC is a spatial simulator of biochemical networks. It supports multiple regions with arbitrarily shapes specified using Constructive Solid Geometry (CSG). SSC also produces fast simulations by directly generating machine code tailored to a specific architecture [1].
Description
BioNetGen to SSC translator provides a SSC equivalent model to your BNG model. The translator outputs two files, one with the translated rules, as modelName.rxn, and the other containing the definition of variables in them, as modelName.cfg. The commands used to generate .rxn and .cfg files are writeSSC and writeSSCcfg respectively. However, there are some features of BioNetGen, which are not supported in SSC. For a description of these see Release Notes. Also, for other debugging issues see An introduction to BNG to SSC translator and the SSC Reference Manual [1].
Downloading
BNG to SSC translator is included in BionetGen versions 2.1.6 and up. Click here to download
Getting Started
BNG and SSC are two different languages, semantically and syntactically. Starting from the keywords to the implementation of each reaction rule, the following table briefs the file structure of BNG and SSC, .bngl and .rxn respectively.
| BNG | SSC |
|---|---|
| File extension: .bngl | File extension: .rxn |
| Begin parameters
... End parameters | NA
But can be supplied as an external .cfg file. |
| Begin molecule types
... End molecule types | NA |
| NA | CSG Definition; Eg
region World box width 1 height 1 depth 1 subvolume edge 1 |
Begin seed species A() 100 End seed species | New<molecule name> at conc
new A() at 100 |
Begin observables Molecules A_tot A() End observables | Record <molecule,specie,pattern name>
Record A() |
Begin reaction rules A() + B() -> C() k1 End reaction rules | Rxn pattern at propensity -> actions
rxn Mol1:A() + Mol2:B() at k1 -> destroy complex Mol1; destroy complex Mol2; new C() |
| actions block
generate_network; simulate_ode; simulate_ssa writeSSC | These ations are performed outside the
.rxn file |
Please note that, as in Table 1, Mol1, Mol2 and Moln, (where n is number of molecules involved) is a BNG-SSC translator keyword. They will be referred as labels of the pattern or molecule. Any operation in SSC except, create new molecule and breaking of bonds, is defined with the help of these labels. Also, in SSC the scope of, labels of patterns and molecules, is local and it will only be placed in front of the molecule on which operation is been performed.
Table 2, shows various equivalents and non-equivalents of SSC in BNG
| BNG term | SSC equivalent |
|---|---|
| Bond symbol ! | Bond symbol # |
Component state defined as ~<state>
A(a~U) | Component state defined as ="<state>"
A(a="U") |
Allows ~<state>!1
A(a~U!1) | Allows, but as <component name> = "<state>"; <component name>Binds#1
A(a="U";aBinds#1) |
Bound molecules are separated by .
A(a!1).B(b!1) | Bound molecules are separated by nothing
A(a#1)B(b#1) |
| + in reactant side, for two or more reactants | <space>, for two or more reactants |
| + in product side, for two or more products | ;, for two or more product operation. (syntax doesn't list products) |
| <->, allows reversible reactions | reversible reactions are expanded as two. The translator can expand such a rule. |
| +, for pattern matching of a bound component | _, underscore symbol |
| Trash() or when deleting an entire specie | Destroy complex <patern,specie,molecule label> |
| DeleteMoleules or when there is any molecule still present
in the reactant side A().B()->A() k1 DeleteMolecules A().B() -> A() k1 | Destroy <pattern,specie,molecule label>
rxn A() Mol1:B() at k1 -> destroy Mol1 rxn A() Mol1:B() at k1 -> destroy Mol1 |
| Comments as, # | Comments as, --# |
Instructions for using BNG to SSC translator can be found here.
Example
Let us consider a simple Gene Expression model from van Zon and ten Wolde., 2006[2]. For more understanding see Reference [2].
| DNA + A | => | DNA.A | Ka | ;Ka | =3e9/M/s |
| DNA.A | => | DNA + A | Kd | ;Kd | =21.5/s |
| DNA.A | => | P + DNA + A | Kprod | ;Kprod | =89.55/s |
| P | => | decay | Kdec | ;Kdec | =0.04/s |
Initially there is one free DNA site fixed in the center and 18 molecules A (corresponds to a 30nm concentration) diffusing with D=1μm2s-1.
We will write this as geneExpr.bngl.
begin parameters
ka 5
kd 21.5
kprod 89.55
kdec 0.04
DNA_init 1
A_init 18
end parameters
begin molecule types
DNA(a)
A(dna)
P
TrashP
end molecule types
begin seed species
DNA(a) DNA_init
A(dna) A_init
P 0
TrashP 0
end seed species
begin observables
Molecules Protein P
end observables
begin reaction rules
Rule 1 A binds to DNA: DNA(a)+A(dna) <-> DNA(a!1).A(dna!1) ka,kd
Rule 2 P synthesis: DNA(a!1).A(dna!1) -> P + DNA(a) + A(dna) kprod
Rule 3 P Degrade: P -> TrashP kdec
end reaction rules
generate_network({overwrite=>1});
simulate_ssa({t_end=>1000,n_steps=>99});
writeSSC;
writeSSCcfg;
Note:- ka is a bimolecular rate constant, divide it by Avogadro's number * Volume
This outputs the following two files:
- geneExpr.rxn
--# SSC-file for model geneExpr created by BioNetGen 2.1.6 region World box width 1 height 1 depth 1 subvolume edge 1 --# Initial molecules and their concentrations new DNA() at DNA_init new A() at A_init new P() at 0 new TrashP() at 0 --# reads observables record P() --# reaction rules rxn Mol1:DNA(a#) Mol2:A(dna#) at ka -> Mol1.a#Mol2.dna rxn DNA(a#1)A(dna#1) at kd -> break 1 rxn DNA(a#1)A(dna#1) at kprod -> new P(); break 1 rxn Mol1:P() at kdec -> destroy complex Mol1; new TrashP()
To the .rxn file apply the following modifications:
- subvolume edge is given as 1. This is well mixed. For heterogeneous system:
subvolume edge 0.1
- There is no information about diffusion. This, users have to add on their own as per their needs. We define parameter D in the .cfg file.
diffusion A(dna#) at D
- geneExpr.cfg
# SSC cfg file for model geneExpr created by BioNetGen 2.1.6 # begin parameters ka = 5 kd = 21.5 kprod = 89.55 kdec = 0.04 DNA_init = 1 A_init = 18 # end parameters
To the .cfg file apply the following modifications:
- Divide the bimolecular rate constants with (subvolume edge)3 and diffusion rate constants with (subvolume edge)2.
ka = 5e3 D = 100
Result
After running the BNGL model and its equivalent SSC model as output by the translator we get the following result:
Thus, we see how oscillations are introduced due to spatial effects. By making our model spatial we are incorporating correct biological environment that as DNA is fixed in one of the compartments, and A is freely diffusing in other compartments, A takes time to find DNA and bind to it. This causes fluctuation in the protein concentration.
References
- Mieszko Lis, Maxim N. Artyomov, Srinivas Devadas, and Arup K. Chakraborty. Efficient stochastic simulation of reaction–diffusion processes via direct compilation. Bioinformatics 2009 25: 2289-2291.
- Maciej Dobrzynski, Jordi Vidal Rodríguez, Jaap A. Kaandorp, and Joke G. Blom. Computational methods for diffusion-influenced biochemical reactions. Bioinformatics 2007 23: 1969-1977.
