# Team:UCL/Science/Model

### From 2014.igem.org

Line 64: | Line 64: | ||

<BR> <BR> | <BR> <BR> | ||

<p> Approximate Bayesian Computation (ABC) is a method that utilises Bayesian statistics for parameter inference in synthetic biology. Given a model and data form that model, it computes the most likely parameters that could give rise to that data. We used the model and simulated data we had in order to find out which parameters are restricted in the values they can have in order to achieve that behaviour. </p> | <p> Approximate Bayesian Computation (ABC) is a method that utilises Bayesian statistics for parameter inference in synthetic biology. Given a model and data form that model, it computes the most likely parameters that could give rise to that data. We used the model and simulated data we had in order to find out which parameters are restricted in the values they can have in order to achieve that behaviour. </p> | ||

- | <p> To use ABC-SysBio we had to make an SBML file describing our model and write an xml input file. The input file contains values for initial conditions of each species in our model, as well as prior distributions for each parameter. The prior distributions consist of a range of values for each parameter, from which the algorithm will sample values. The input file also contains the data from the degradation of sulfonated AzoDyes over two days. </p> | + | <p> To use ABC-SysBio we had to make an SBML file describing our model and write an xml input file. The input file contains values for initial conditions of each species in our model, as well as prior distributions for each parameter. The prior distributions consist of a range of values for each parameter, from which the algorithm will sample values. The input file also contains the data from the degradation of AzoDyes and sulfonated AzoDyes over two days. </p> |

<BR> <BR> | <BR> <BR> | ||

<p>ABC-SysBio samples a value for each parameter from the priors and using the initial conditions provided, simulates the model. The resulting time course is compared to the data provided, and if the distance between the two is greater than a threshold, the sampled parameter set is rejected. This is repeated for 100 sets of samples, consisting of one population. The sets that were accepted are then perturbed by a small amount and then a new population is sampled from the perturbed sets. This process is repeated until the distance between the data and the simulations is minimised. The parameter values that gave rise to this final population are called the 'posterior distribution'. This is shown in the figure below: </p> | <p>ABC-SysBio samples a value for each parameter from the priors and using the initial conditions provided, simulates the model. The resulting time course is compared to the data provided, and if the distance between the two is greater than a threshold, the sampled parameter set is rejected. This is repeated for 100 sets of samples, consisting of one population. The sets that were accepted are then perturbed by a small amount and then a new population is sampled from the perturbed sets. This process is repeated until the distance between the data and the simulations is minimised. The parameter values that gave rise to this final population are called the 'posterior distribution'. This is shown in the figure below: </p> | ||

<BR> <BR> | <BR> <BR> | ||

- | <p>The distribution of values for each parameter are shown in the diagonal. All distributions are between 0 and 1. Drawing a straight line from one parameter to the other, at the point where the two meet, the two parameters have been plotted against each other in a density contour plot. Three parameters stand out as very constricted, k3, k17 and k18. These are the parameters of the reactions for intake | + | <p>The distribution of values for each parameter are shown in the diagonal. All distributions are between 0 and 1. Drawing a straight line from one parameter to the other, at the point where the two meet, the two parameters have been plotted against each other in a density contour plot. Three parameters stand out as very constricted, k3, k17 and k18. These are the parameters of the reactions for intake (k3) and secretion (k8) of AzoDyes as well as the intake (k17) and secretion (k18) of sulfonated AzoDyes by the cell. This shows that the bottleneck happens at those points in our pathway. So if we increase the rate of intake and secretion of AzoDyes and sulfonated AzoDyes in our synthetic pathway, we could speed up the process of degradation! </p> |

<li>Posterior distribution of model parameters</li> | <li>Posterior distribution of model parameters</li> | ||

<img src="http://2014.igem.org/wiki/images/7/7f/Azo_posterior.png" class="imgsizecorrect"> | <img src="http://2014.igem.org/wiki/images/7/7f/Azo_posterior.png" class="imgsizecorrect"> |

## Revision as of 21:26, 17 October 2014

- Modelling Degradation
- Parameter Inference
- Flux Balance Analysis
- Enzyme Kinetics
- Chemical Mechanism
- References

### Overview

There are three ways we can degrade azodyes: using Azoreductase (AzoR), Laccase (Lac) or BsDyp. Azoreductase breaks down AzoDye (AzoD) into two products Laccase breaks down AzoDye as well as the products of the reaction of Azoreductase with AzoDye. BsDyP acts on sulfonated AzoDyes (sAzoD):

In order to model this system we used COPASI. We included equations for gene expression and degradation for each gene in our pathway, as well as the intake and excretion of AzoDyes and sulfonated AzoDyes. The equations we included as well as the parameter assigned to each one are shown below:

**The Role of Microfluidic Analysis to Evaluate the Scalable Synbio Azo-Remediation Solution**

**designed and tested**a novel approach to azo-remediation, which allows sustainable and scalable bioprocessing. Our bioprocess integrates elements from

**upstream**and

**downstream**processing.

In order to develop and improve the functionality of our bioprocess, key steps must be tested to quantify

**process variables**, and allow for preliminary mass transfer calculations and detection of azo dye degradation rates.

We have created microfluidic prototype devices to test the mixing in our reactors, and to test the performance of our novel immobilisation module, allowing for process optimisation and testing, without the

**burdens**of expensive pilot scale testing.

The process testing timeline demonstrates that effective microfluidic testing can be used in replacement to conventional small-scale testing approaches. This is ideal for our project, especially when optimising whole unit operations.

Using reasonable parameter values, the simulation showed that the AzoDye is degraded within two days (48 hours). This timeframe agrees with the experimental results!

### Parameter inference

We wanted to see which part of the pathway is the bottleneck in degrading the sulfonated AzoDyes. So we analysed the parameters of our model to see which one is the most constrained, which could give us an insight on which one to tweak experimentally in the future in order to speed up the degradation. To do that we used ABC-SysBio (Liepe, 2014) .

Approximate Bayesian Computation (ABC) is a method that utilises Bayesian statistics for parameter inference in synthetic biology. Given a model and data form that model, it computes the most likely parameters that could give rise to that data. We used the model and simulated data we had in order to find out which parameters are restricted in the values they can have in order to achieve that behaviour.

To use ABC-SysBio we had to make an SBML file describing our model and write an xml input file. The input file contains values for initial conditions of each species in our model, as well as prior distributions for each parameter. The prior distributions consist of a range of values for each parameter, from which the algorithm will sample values. The input file also contains the data from the degradation of AzoDyes and sulfonated AzoDyes over two days.

ABC-SysBio samples a value for each parameter from the priors and using the initial conditions provided, simulates the model. The resulting time course is compared to the data provided, and if the distance between the two is greater than a threshold, the sampled parameter set is rejected. This is repeated for 100 sets of samples, consisting of one population. The sets that were accepted are then perturbed by a small amount and then a new population is sampled from the perturbed sets. This process is repeated until the distance between the data and the simulations is minimised. The parameter values that gave rise to this final population are called the 'posterior distribution'. This is shown in the figure below:

The distribution of values for each parameter are shown in the diagonal. All distributions are between 0 and 1. Drawing a straight line from one parameter to the other, at the point where the two meet, the two parameters have been plotted against each other in a density contour plot. Three parameters stand out as very constricted, k3, k17 and k18. These are the parameters of the reactions for intake (k3) and secretion (k8) of AzoDyes as well as the intake (k17) and secretion (k18) of sulfonated AzoDyes by the cell. This shows that the bottleneck happens at those points in our pathway. So if we increase the rate of intake and secretion of AzoDyes and sulfonated AzoDyes in our synthetic pathway, we could speed up the process of degradation!

### Flux Balance Analysis

In order to see whether our xenobiological approach would work we wanted to check whether lack of Ubiquinone would have an effect on the growth rate of the chassis. The literature (Okada 1997 and Soballe, 1999) suggested that Ubiquinone is essential for E.coli growth so we decided to put that to the test! In order to do that we used Flux Balance Analysis (FBA). FBA is a method that uses the metabolism model of E.coli (see below) and calculates the flow of metabolites through that system that is required to maximise a given obective

Ecoli metabolism plotted in Cytoscape (Cline, 2007): In our case we used growth rate as the objective to maximise. We performed FBA for the core E.coli metabolism with and without Ubiquinone present. With Ubiquinone present the growth rate was calculated to be 0.98 h^{ -1 }. Without Ubiquione in the system the growth rate was found to be 0 h^{ -1 }, indicating that E.coli would not grow and survive without ubiquione. This suggested that silencing the essential genes for Ubiquinone production and supplying it externally would give us control over the survival of the chassis and ultimately allow us to contain it.

Core metabolism map used for FBA

### References

Liepe, J., Kirk, P., Filippi, S., Toni, T., et al. (2014) A framework for parameter estimation and model selection from experimental data in systems biology using approximate Bayesian computation. [Online] 9 (2), 439–456.

Hoops S., Sahle S., Gauges R., Lee C., Pahle J., Simus N., Singhal M., Xu L., Mendes P. and Kummer U. (2006). COPASI: a COmplex PAthway SImulator. Bioinformatics 22, 3067-74.

Cline, M.S., Smoot, M., Cerami, E., Kuchinsky, A., et al. (2007) Integration of biological networks and gene expression data using Cytoscape. Nature Protocols. [Online] 2 (10), 2366–2382.

Orth, J.D., Thiele, I. & Palsson, B.O. (2010) What is flux balance analysis? Nature Biotechnology. [Online] 28 (3), 245–248.

Okada, K., Minehira, M., Zhu, X., Suzuki, K., et al. (1997) The ispB gene encoding octaprenyl diphosphate synthase is essential for growth of Escherichia coli. Journal of bacteriology. 179 (9), 3058–3060.

Søballe, B. & Poole, R.K. (1999) Microbial ubiquinones: multiple roles in respiration, gene regulation and oxidative stress management. Microbiology (Reading, England). 145 ( Pt 8)1817–1830