Collaborations

Meetups and discussions

Throughout the time of iGEM, we have participated in multiple activities to get in touch with other teams! This was a great opportunity for us to exchange, get some advice and learn from each other.
The iGEM European meetup, hosted by the team of Hamburg was the perfect opportunity to do this! In the course of three days, we met many different teams from all over Europe. In workshops and poster sessions, we could give each other very helpful pieces of advice about how to present a project in an exciting, understandable way. This event was a great possibility to discuss not only one's project but also workflow, organization and plans for the future. By sharing experiences, thoughts and ideas, we could all learn from the successes but also setbacks of other teams. On top of that, we could network with a great number of people, and of course, we found some new friends!

Embo seminar

As a part of our human practices efforts, we have organized a seminar on the topic of research integrity.
We had the honour of inviting Michele Garfinkel as a speaker, an Embo associate that has worked with the truthful representation of research on a daily basis. This seminar was targeted towards fellow iGEM teams, full of young researchers, upon many of which were confronted with research integrity for the first time during iGEM. It was an incredibly great opportunity for all of us to learn about this crucial topic.
Additionally, after receiving great input by Michele, we could discuss interesting ideas together with the other teams, reaching from the reasons for the falsification of data, possible problems emerging from the pressure on scientists to publish, over possible solutions to this, to important aspects we as iGEM teams should look out for to produce research that is as honest, transparent and reliable as possible! If you want to learn more about our seminar, click here to visit our integrated human practices page!

Modelling with Cambridge

Collaboration with team Cambridge. chAMBER meets AdaptR - the beginning of a new chAptR

For many aspects of our project measuring bacterial growth curves plays an important role. To deepen our understanding of the growth process observed during our assays we decided to model our data. After taking part in iGEM’s workshop “Modeling and Analysis of Synthetic Biology Systems with SimBiology and MATLAB” we decided to try modelling and curve-fitting of our OD measurements for bacterial growth with MATLAB.  

We soon realized that this was the perfect opportunity to collaborate with another iGEM team. Growth curve data will be acquired in many different projects, and the workshop organized by iGEM helped us connect with other teams interested in modelling with MATLAB.  

Our goal for this collaboration was to evaluate our modelling approaches and eliminate potential bias during curve fitting. We hoped to broaden our horizon by engaging and discussing with colleagues from the scientific community and help us to see our data from different angles.  

We got in contact with the Cambridge iGEM team and discussed the possibilities of our potential collaboration. We had a very productive brainstorming meeting to discuss the ramification of our collaboration, and we agreed to exchange the raw growth curve data with each other. Each team would then model the other team's data, as well as its own data using an approach of its choice. After completing the modelling, we arranged a follow-up meeting and compared and discussed our results and strategies. 

Figure 1: Collaboration meeting with team Cambridge. Joni Wildman (top left) and Shau'ri Wiggins (bottom left) discussed their modelling approach with us. 

Model comparison 

The following section will show a few selected examples to illustrate our modelling process.  Cambridge gave us data in the form of OD600 measurements of bacterial growth curves under different conditions and fluorescence intensity measurements for the same conditions. We provided them with similar data, OD600 measurements of production assays in E.coli BL21 (DE3). Both teams used the MATLAB curve fitting tool. 

We decided to model the bacterial growth with a modified version of the logistic model, which was adapted from M. H. Zwietering et al.[1] and M. Kahm et al. [2]. The parameter λ describes the lag-phase, A is the asymptotic value, μ stand for the specific growth rate and c is the offset. 

Team Cambridge decided on a slightly different approach by using the differential equation describing logistic growth. They calculated dN/dt, the rate of change of growth at each time point and fit the following parabola equation. Here, a represents the maximum growth rate, and b represents the carrying capacity of the population. 

We compared the parameter of the fits and the goodness of fit values for our data. The MATLAB curve fitting tool provides multiple parameters to characterise curve fits: 

• The SSE measures the total deviation of the response values from the fit to the response values. It is also called the summed square of residuals and is usually labelled as SSE. A value closer to 0 indicates that the model has a smaller random error component and that the fit will be more useful for prediction. 
The R-square 
•  value is the proportion of the variation in the dependent variable that is predictable from the independent variable(s). 
• The adjusted R-square value can be interpreted as a less biased estimator of the population R-square, whereas the observed sample R-square is a positively biased estimate of the population value. Adjusted R-square is more appropriate when evaluating model fit (the variance in the dependent variable accounted for by the independent variables) and in comparing alternative models 
• The RMSE (root mean squared error) is the square root of the variance of the residuals. It indicates the absolute fit of the model to the data–how close the observed data points are to the model’s predicted values. Whereas R-squared is a relative measure of fit, RMSE is an absolute measure of fit. A lower RMSE indicates a better fit. 

 An example for our difference in fits is the data from a growth curve of BL21 at 37°C from our lab.  

In Figure 2 our fit for the growth curve is shown. 

Figure 2: This is a growth curve for E.coli BL21 at 37°C in LB medium shown as blue dots. The fit model shown in red is described by a modified logistic model: y(t) = A/(1+exp((4μ/A)(μ-t)+2))-c 

The fit model we received from Cambridge is shown in Figure 3. 

Figure 3: This is based on a growth curve for E.coli BL21 at 37°C in LB medium. The blue dots represent the rate of change in OD plotted against the total growth. The fit model shown in red is described by a logistic model: dN/dt = a((b-N)/b)N 

Table 1 sums up the fit parameters used by us and our collaboration partners.  

Table 1: Comparison of fit models describing bacterial growth curves used for curve fitting with the MATLAB curve fitting tool.

The models performed very differently. It is striking, that Cambridges model scores lower for R-square and associated values, not only in this example but in fits for other growth curves as well. We discussed some ideas about why that could be and what makes our models behave differently.  

The model for logistic growth assumes stable growth conditions. In real-life experiments, these conditions are rarely met. Especially when measuring OD600 in a plate reader. Here the bacterial cells are bound to run out of nutrients at some point, which will influence their growth behaviour. After they have reached their stationary phase, they will certainly enter the death phase soon. Growth conditions in a well-plate are heavily restricting in this aspect. Therefore, we suggest excluding data from the death phase from the fits, as the logistic model does not take cell death into account and the OD measurements will decrease over time. 

Additionally, we suggest using an off set in all fits. Again, the standard logistic model assumes perfect starting conditions, which would mean the data starts at zero. However, in real measurements the equipment can cause off sets or samples have a certain background. To adjust your fitting functions to this it is enough to add a constant c. So the original function dN/dt=a*((b-x)/b)*x becomes dN/dt=a*((b-x)/b)*x+c and will more realistically model the data. It is also possible to calculate the off-set from the control samples, so no additional variable is necessary. This means adding the mean of a negative control.  

Conclusion 

We learned a lot discussing our different approaches and learning about the way other research groups process their data. The collaboration helped us identify anomalous results by getting an unbiased opinion. Showing the raw data of our experiments improved transparencies between our teams. This collaboration helped both our teams to improve, accelerate and enhance our data analysis process. 

We also really enjoyed getting to know other iGEMers! Being part of the synthetic biology community and getting their feedback is invaluable and pushed us personally forward. 

The Transcriptome, a blog from iGEM UNILausanne and Chalmers-Gothenburg

During our project, we have realized that science communication is a crucial field that is of incredible importance right now more than ever. Because of that, we found great interest in collaborating with the iGEM teams Chalmers-Gothenburg and Lausanne. Their teams have a partnership reaching back to 2020, when they together created the blog “The Transcriptome”.
Each week, a new article is posted on their website, explaining exciting concepts, applications, and achievements in the field of synthetic biology! By structuring these articles in a meaningful, yet understandable and interesting way, people foreign to the scientific bubble are invited to learn something about SynBio. A core idea is that this blog should not only be interesting, but also accessible for as many people as possible.
That is exactly where we started. First, we agreed to translate "How biotechnologies facilitate the production of insulin" and "Extinguishing malaria with genetic engineering" into German. This was a lot of fun, so we decided to start translating the whole website into German because the German infrastructure was missing. After getting back to Chalmers-Gothenburg and UNILausanne with the translation of the website, we had a call for talking about a bigger collaboration. They asked us if we also wanted to write articles ourselves and of course we could not refuse. Once all the red tape was sorted out, we got access to the website and started building the infrastructure for the German website first, so that it would be as understandable for German readers as it is for readers of other languages.
Now the page is fully accessible in German! Finally, we have agreed to continuously translate upcoming articles and even contribute by writing some ourselves! In teams consisting of one person from UNILausanne / Chalmers-Gothenburg and the other being from our team, we have written four articles in total (2 of which are published after the wiki freeze)! We gathered information for the articles we wanted to write. Lucas Toftås and Michael Spädt collaborated on the article "GMOs And Purple Tomatoes". Leon-Samuel Icking has written an article with Emil Löfgren on vitamin A deficiency and Golden Rice. In it, they describe the serious consequences that a vitamin A deficiency can have, but also a solution, Golden Rice, which could eliminate this deficiency and how it does it. Anders Källberg and Leon-Samuel Icking collaborated on an article on AlphaFold to explain what the problems are in these protein folding calculations and how they were solved.
We were incredibly honoured when we were asked to be included on the roster page “people behind the posts” as authors together with our lads from UNILausanne and Chalmers-Gothenburg!
This awesome blog was an awesome experience and great opportunity for all of us to exchange interesting thoughts about synthetic biology while being able to share our ideas with a wider audience. It was a bliss to cooperate with these teams to achieve our common goal of educating, inspiring and including people into science in an accessible and creative way, that was on top of all a lot of fun for us as well as (hopefully) for our readers!