Latest revision as of 17:43, 29 October 2008

Newcastle University

GOLD MEDAL WINNER 2008

Home	Team	Original Aims	Software	Modelling	Proof of Concept Brick	Wet Lab	Conclusions

Home >> Original Aims

Original Aims

Bioinformatics is essential to modern biology, but is often considered as merely a tool, like a word processor or a spreadsheet, rather than an important approach to understanding and solving biological problems.

We consider bioinformatics to be not only a tool in biology, but as a viable method for exploring a problem. Wet-lab biology is expensive in terms of time, money, and manpower. A single bioinformatician can investigate a given problem with no equipment other than a computer, and run many iterations of the same experiment within seconds, rather than the weeks that the labs may take.

However, the bioinformatics is only as good as its data and simulation power. The field expands daily with new information, all of which must be incorporated into a simulation in order for it to give useful results. Much of bioinformatics is backed by hard data from wet labs. Running an experiment 1,000 times is only useful if you know what all the variables are, and what they should be.

The most sensible approach to the biological investigations is to take advantahe both of of these different yet complementary methods - bioinformatics and wet lab experimentation - taking advantage of the strengths of each, while hopefully minimizing the limitations.

BugBuster

We aimed to develop a diagnostic biosensor for detecting pathogens. We wanted this to be cheaply and readily available for deployment in areas with limited medical resources such as refrigeration and sophisticated laboratories. We chose to use Bacillus subtilis as a method of delivery due to its ability to sporulate. The sensor bacteria could be dried down as spores, which are extremely resilient to environmental conditions, and could be rehydrated as required.

Our aim was to develop a strain of Bacillis subtilis 168 with the ability to detect four Gram-positive pathogens through their extracellularly-secreted quorum-sensing peptides, and to indicate through reporter genes linked to the receptor-ligand cascade which of these peptides are in its growth medium.

Gram-positive bacteria communicate using quorum communication peptides. Research has shown that they are extremely strain-specific. B. subtilis has a range of genes that enable detection of these peptides. The issue that we had to overcome is the fact that we have more quorum-sensing peptides to detect than fluorescent proteins to show their presence. The output of the system is followed by analysing expression of fluorescent proteins such as mCherry, GFP, CFP and YFP.

Mapping multiple inputs to three output states is a multiplexing problem. The design of the genetic circuitry to do this is non-trivial and is not feasible manually. We therefore chose to use a biological implementation of an artificial neural network (ANN). Our team members wrote, designed and implemented a complete suite of tools that allowed the design and simulation of regulatory networks. An essential part of the approach was the use of computational evolution to design circuits with predictable behaviour even when the details of the required topology are unknown in advance. We used the modelling language CellML[http://www.cellml.org/] because of its ability to easily model virtual parts that can be easily assembled into a circuit which can be simulated. We aimed to translate this model into a sequence that could be implemented as one or more BioBricks.

An overview of our complete system.

A visual output was chosen for detection of pathogens, as this allows for rapid detection without a requirement for specialist equipment. The aim was to have different fluorescent protein outputs turned on by the presence of different pathogenic bacteria, and different combinations of these bacteria. In line with the neural network concept we wanted to map numerous inputs to limited outputs. Initially, we aimed to detect four pathogens, and we wanted different outputs for each of these and each combination of these.

We aimed to produce a workbench that incorporates a parts repository, constraints repository and an evolutionary algorithm (EA). The EA takes input from the parts repository and constraints repository and evolves a neural network, using the CellML modelling language to carry out simulations of the network behaviour, which can be assessed for fitness (how closely the desired behaviour is reproduced). The fittest model can then be used to generate a DNA sequence implementing the neural network in vivo. This DNA sequence can then be synthesized and cloned into the B. subtilis chassis. One of our outcomes should be a range of neural network node BioBrick devices which can be combined to form the in vivo neural network.

In keeping with the neural network model, our bacterial circuit would have three layers of complexity. The plan can be summed up as follows:

The input layer to the bacterial neural network is represented by the two-component genes, activated by the peptides from the four gram-positive pathogens. Each peptide represents a node.
The hidden layer is represented by an assortment of different transcription factors.
The output layer is represented by three fluorescent proteins; GFP, YFP and mCherry.

The biologist specifies the inputs and the outputs. In the presence of a particular profile of peptides, a specific fluorescent protein, or combination of proteins, will be expressed.

This was the starting point from which the construct was designed.

@@ Line 1: / Line 1: @@
-<html>
-<style type="text/css">
-#maincol {
-	 width: 65%;
-top:0em;
-}
-</style>
-</html>
 {{:Team:Newcastle University/Header}}
-{{:Team:Newcastle University/Template:UnderTheHome|page-title=[[Team:Newcastle University/Parallel Evolution|Approach]]}}
+{{:Team:Newcastle University/Template:UnderTheHome|page-title=[[Team:Newcastle University/Original Aims|Original Aims]]}}
-{{:Team:Newcastle University/ApproachSideBar}}
+== Original Aims ==
-<div id="maincol">
-== Parallel Evolution ==
-There standard practice in biology today, synthetic or not, is to treat bioinformatics as any other tool to achieve an end. A problem is determined, and the biologist works out a possible avenue of exploration. Sometimes, this involves the use of bioinformatics tools, such as BLAST searches or phenotypic trees. Once the method of exploration is established, the biologist rarely goes back to bioinformatics approaches to analyze her results.
+Bioinformatics is essential to modern biology, but is often considered as merely a tool, like a word processor or a spreadsheet, rather than an important approach to understanding and solving biological problems.
-The concept of parallel evolution treats bioinformatics not only as a tool in biology, but as a viable but limited method for exploring a problem. Wet-lab biology is expensive in terms of time, money, and manpower. A single bioinformatician can test the same situation with no equipment other than a computer, and run many iterations of the same experiment within seconds, rather than the weeks that the labs may take.
+We consider bioinformatics to be not only a tool in biology, but as a viable method for exploring a problem. Wet-lab biology is expensive in terms of time, money, and manpower. A single bioinformatician can investigate a given problem with no equipment other than a computer, and run many iterations of the same experiment within seconds, rather than the weeks that the labs may take.
-However, the bioinformatics is only as good as its simulation. The field expands daily with new information, all of which must be incorporated into the simulation in order for it to give useful results. Much of bioinformatics is backed by hard data from wetlabs. Running an experiment 1000 times is only useful if you know what all the variable are, and what they should be.
+However, the bioinformatics is only as good as its data and simulation power. The field expands daily with new information, all of which must be incorporated into a simulation in order for it to give useful results. Much of bioinformatics is backed by hard data from wet labs. Running an experiment 1,000 times is only useful if you know what all the variables are, and what they should be.
-The most sensible approach to the problem of these different yet complementary methods is to play one off of the other, taking advantage of the strengths while minimizing the limitations.
+The most sensible approach to the biological investigations is to take advantahe both of of these different yet complementary methods - bioinformatics and wet lab experimentation - taking advantage of the strengths of each, while hopefully minimizing the limitations.
-</div>
+==BugBuster==
+We aimed to develop a diagnostic biosensor for detecting pathogens. We wanted this to be cheaply and readily available for deployment in areas with limited medical resources such as refrigeration and sophisticated laboratories. We chose to use ''Bacillus subtilis'' as a method of delivery due to its ability to sporulate. The sensor bacteria could be dried down as spores, which are extremely resilient to environmental conditions, and could be rehydrated as required.
-== Drylab Approach ==
+Our aim was to develop a strain of [[Team:Newcastle University/Bacillus subtilis|''Bacillis subtilis 168'']] with the ability to detect four Gram-positive pathogens through their extracellularly-secreted quorum-sensing peptides, and to indicate through reporter genes linked to the receptor-ligand cascade which of these peptides are in its growth medium.
-The two techniques we decided to use in our bioinformatics approach were neural networks and genetic algorithms. Computer science has consistently looked to Mother Nature for inspiration, and the results are best exemplified with these two techniques.
+Gram-positive bacteria communicate using quorum communication peptides. Research has shown that they are extremely strain-specific. ''B. subtilis'' has a range of genes that enable detection of these peptides. The issue that we had to overcome is the fact that we have more quorum-sensing peptides to detect than fluorescent proteins to show their presence. The output of the system is followed by analysing expression of fluorescent proteins such as mCherry, GFP, CFP and YFP.
-[[Image:Neuron.png|right|thumbnail|A quick sketch of the anatomy and connection of neurons]]
+Mapping multiple inputs to three output states is a multiplexing problem. The design of the genetic circuitry to do this is non-trivial and is not feasible manually. We therefore chose to use a biological implementation of an artificial neural network (ANN). Our team members wrote, designed and implemented a complete suite of tools that allowed the design and simulation of regulatory networks. An essential part of the approach was the use of computational evolution to design circuits with predictable behaviour even when the details of the required topology are unknown in advance. We used the modelling language CellML[http://www.cellml.org/] because of its ability to easily model virtual parts that can be easily assembled into a circuit which can be simulated. We aimed to translate this model into a sequence that could be implemented as one or more BioBricks.
-Neural networks are a method of machine learning based on neurons in the brain. In a neuron, signals are received at the dendrites. If the combined triggering impulse is higher than a certain threshold, the neuron fires. The impulse travels down the cell membrane of the axon and is transmitted via the axon terminators to the dendrites of the next neuron. When the neuronic pathway is traveled often, the connections are reinforced. The axon of the preceeding neuron and the dendrites of the subsequent neuron grow to be more highly connected. It is in this way that we learn. and why we can change behaviour patterns once they are established. Forcing the impulse to take another route results in an altered behaviour or thought.
-This same approach is used to teach a computer how to 'think' about a problem. The neurons in the system become nodes. The nodes are arranged in several layers. "Impulses" begin at the input layer, travel through any number of hidden layers, and terminate at the output layer. The output layer gives the computer its learned behaviour.
+[[Image:Overview.GIF|500px|center]]
-Nodes can be highly interconnected. The connections themselves have a weight, which corresponds to the number of axon-dendrite connections in neurons. Nodes that are connected with a heigher weight are more likely to fire.
+<center>An overview of our complete system.</center>
-Neural networks must be trained in order to function properly. Untrained neural networks are like newborn babies: cute, but unable to perform any higher functions. They must be taught.
+A visual output was chosen for detection of pathogens, as this allows for rapid detection without a requirement for specialist equipment.  The aim was to have different fluorescent protein outputs turned on by the presence of different pathogenic bacteria, and different combinations of these bacteria. In line with the neural network concept we wanted to map numerous inputs to limited outputs.  Initially, we aimed to detect four pathogens, and we wanted different outputs for each of these and each combination of these.
-==Wetlab Approach==
+We aimed to produce a [[Team:Newcastle University/Workbench|workbench]] that incorporates a [[Team:Newcastle University/Parts Repository|parts repository]], [[Team:Newcastle University/Constraints Repository|constraints repository]] and an [[Team:Newcastle University/Evolutionary Algorithm|evolutionary algorithm]] (EA). The EA takes input from the parts repository and constraints repository and evolves a neural network, using the CellML modelling language to carry out simulations of the network behaviour, which can be assessed for fitness (how closely the desired behaviour is reproduced). The fittest model can then be used to generate a DNA sequence implementing the neural network ''in vivo''. This DNA sequence can then be synthesized and cloned into the ''B. subtilis'' chassis. One of our outcomes should be a range of neural network node BioBrick devices which can be combined to form the ''in vivo'' neural network.
-Newcastle’s 2008 iGEM team is aiming to transform plasmids grown in ''Escherichia coli'' into ''Bacillis subtilis'' (a gram-positive bacterium), with the aim of integrating this stably into the ''B. subtilis'' chromosomal DNA. By doing this, we hope to create an organism with the ability to detect a range of extracellularly secreted quorum-sensing peptides, and indicate through reporter genes linked to the receptor-ligand cascade which of these are in its growth medium.
+In keeping with the neural network model, our bacterial circuit would have three layers of complexity. The plan can be summed up as follows:
-[[Team:Newcastle University/Bacillus subtilis|About Bacillus subtilis]]
+* The input layer to the bacterial neural network is represented by the two-component genes, activated by the peptides from the four gram-positive pathogens. Each peptide represents a node.
+* The hidden layer is represented by an assortment of different transcription factors.
+* The output layer is represented by three fluorescent proteins; GFP, YFP and mCherry.
+The biologist specifies the inputs and the outputs. In the presence of a particular profile of peptides, a specific fluorescent protein, or combination of proteins, will be expressed.
+This was the starting point from which the construct was designed.