Difference between revisions of "Team:TU-Eindhoven/Team/Notebook"

Throughout the past few months we have tracked our progress in our notebook. Below we give a short overview per month including what we have accomplished that month and what kind of experiments we did. For a more detailed overview of the labwork, click on the arrow.

March - Building the Team

During this first month we divided the team roles and we started to brainstorm about project ideas and read a lot of literature for inspiration. Every week we had a meeting to discuss our progress and exchange ideas.

April - Constructing the Constructs

After a lot of brainstorm sessions we started to design our constructs and had meetings with our supervisors to finetune the idea. During the end of the month we started working on designing the DNA of the constructs. We also had a first session for our Human Practices as a preparation for SYNENERGENE at the end of the month.

The first round of gBlocks were designed in such a way that they can be used with the gibson assembly method. We used the cloning guide from iGEM team Eindhoven 2015 as a starting point. For detailed information on our protein design click here.

May - First gBlocks ordered!

After a lot of work, redesign and adding details we ordered our gBlocks! We also got into contact with iGEM Potsdam, to see whether we could collaborate.

June - Prepping the Lab

During the busy month of June we started with some preparatory work, because we had to wait for our gBlocks to arrive. However, towards the end of the month they arrived! For some reason the Gibson Assembly of the 14-3-3 construct did not work, but the Gibson Assembly of the CT33 construct showed promising results.

March - Exploring options

In the first month we divided the roles of the team members, and the team members with experience and interest in modeling were became members of the modeling crew, with the most experienced one as captain. The first tasks were to read literature and brainstorm about way to simulate the system.

April - Contacting Experts

As the idea of the constructs became clearer, it became time to search for people that could advise us about how to start with the model. The paper ‘Compositional Control of Phase-Separated Cellular Bodies’ ^[1] published in Cell by Michael Rosen looked very promising as they used Matlab to simulate their system. We got a very enthusiastic response and they were happy to help us. He forwarded our mail to the person responsible for the model, Salman Rosen, and he send us some functions together with an explanation. The functions were a nice source of inspiration, but unfortunately we had to develop functions of our own.

May - New Insights

In the next period we had conversation with the modeling captains of the TU Eindhoven team of 2016 and 2015 and asked for their experience. We explained what we were thinking about and how we wanted to execute our plan and got some feedback as response. Furthermore, we were inspired by the course “Molecular Modeling” that let us simulate lipids by using a Atomistic Molecular Dynamics Simulation method. Here we had to apply Monte Carlo and Glauber dynamics, of which Monte Carlo also seemed to be useful for our designed constructs.

June - First Attempts

As the lab was preparing for the delivery of the gBlocks, we started with working out the formulas we wanted to use and developed our first model. The first model was based on concentration dependent reaction rates by using an ODE solver. Here we had to define all the possible reactions that could occur in the system. When looking critically to this method, it appeared to be insufficient, as it was impossible to define all the possible reactions.The second attempt included a Molecular Dynamics simulation based on Monte Carlo, followed by investigating the possibilities of a Gillespie algorithm.

July - 2D Glasses

After having some discussions with experts of the Gillespie algorithm, we could conclude that it had the same problem as the ODE solver. This meant we had to continue with expanding the Molecular Dynamics simulation. The first step was to define the molecules and let them move and rotate separately. Also all the other constraints and interaction between the molecules were added, leading to a simulations that showed network formation. The drawback was that it only was a 2D simulation, while in reality it will be in 3D. Additionally, the scale was too low and the constraints needed to many approximations to use it for a useful prediction and verification of the designed system. The simulation yielded a nice short movie, but we had to continue with searching for a different approach, which we found in the BioNetGen Language in combination with NFsim.

August - Virtual Reality

Every (coding) language is different and has its own way of functioning and needs certain requirements. We discovered that when we started with the installation of NFsim, advised by one of our supervisors. He himself hadn’t used it, but knew someone who did, and he needed to use Linux to get it functional. We, having a Windows system, therefore needed to instal a Virtual Box, together with Linux, NFsim and Matlab. This took quite some time and was very frustrating and unrewarding. Especially as we discovered that we could still use Windows by using NFsim somewhat differently. After the installation was successful, we walked through the manual and tried to understand the new language and experimented with example models.

September - Trial and Error

Learning a new language takes some time, but luckily, it was doable in the limited time given by the iGEM deadlines. Of course it took some time to implement the system, but after some trial and error, we ended up with a model that could be simulated with NFsim. However, there were still some problems left, including generating an output in the desired format and having reliable parameters of our system. This led us to dive back into the literature and contact some experts. Both led to success, but then a new problem arose: we only saw a equilibrium state of our system, but no transition form every unbound to a (small) network formation. Different simulation times, steps, rates and so on were tested, but an improvement was not found yet.

October - Enlightening

Units, remember, those are very important!. After doing a unit control, we discovered that we made a small mistake, with large influences on our simulation. So that you know, NFsim works with molecules instead of concentrations. Being used to an ODE solver, we missed the fact that we had to translate the association rate to a molecule dependent value instead of a concentration dependent value. We knew that we were working with molecules instead of concentrations, so we already implemented a volume definition, and together with the number of Avogadro, converted them. After also converting the association rate, the simulation worked much faster. Another insight we gained was that most times, the process is better visualized on a logarithmic time scale. Doing some simulations with different time steps, we could make such a logarithmic time scale plot, and indeed, we saw the transition from the begin state to an equilibrium! Having the simulation working for our system, we could work on simulating the system of the iGEM team from Potsdam.