Navigation and service

Jülich Expertise for American and German Coronavirus Websites

Jülich, 12 August 2020 – Since the start of the COVID-19 outbreak, the main focus has been on the effective reproduction number, also known as the R-value or Rt. It indicates the average number of people an infected person infects with the virus within a specific period of time. In the USA, a website was launched in April, providing the wider public with up-to-date R-value estimates for all US states every day. The website was established by Kevin Systrom and Mike Krieger, who are the co-founders of Instagram. They are supported by open-source software developers, with two doctoral researchers from Jülich’s Biotechnology subinstitute (IBG-1), Laura Helleckes and Michael Osthege, also contributing their expertise to the project. On the website they now provide daily analyses of infection rates in Germany, too.

HelleckesLaura Helleckes
Copyright: Forschungszentrum Jülich / Ralf-Uwe Limbach

What is the main focus of your work at the Biotechnology subinstitute?

Laura Helleckes: We are both doctoral researchers researching the automation of the development of bioproduction processes. Background: There is a large number of organisms to choose from for a process like this, and we can create different operating conditions. For industrial applications, the goal is to find out – based on this range of combinations – what organism is best suited for a particular production process under what conditions. At the institute, hundreds of these growth experiments, which are performed automatically, have to be analysed and transformed into models in a timely manner. We develop the software tools needed for this workflow.

What do these tools look like?

Michael Osthege:  They are statistical tools. I am a member of the development team for the Python software package “PyMC3” for statistical modelling based on Bayesian estimation methods that also include uncertainties. “PyMC3” is an open-source modelling tool that we use at the subinstitute with great efficiency to obtain more information from our experiments.

Michael OsthegeMichael Osthege
Copyright: Forschungszentrum Jülich / Ralf-Uwe Limbach

How did the collaboration with Instagram co-founders Kevin Systrom and Mike Krieger come about?

Michael Osthege: The American website also uses the “PyMC3” package. When the developers were setting up the website, there were technical difficulties with the model. Kevin Systrom reached out to one of the “PyMC3” developers, who in turn got in contact with a few people on the team.

From a mathematical perspective, there isn’t a big difference between analysing microbial growth experiments and analysing epidemiological data. It turned out that our expertise in workflow automation could be transferred at an almost 1:1 ratio to data processing and modelling of the R value. Now, the daily data analysis if fully automated. That means the developers can concentrate on improving the model
We are still in touch but the model is currently running very smoothly. Nevertheless, changes are being made to the programming behind the scenes, which I help out with on occasion..

Can the model also be applied to Germany?

Laura Helleckes: Yes, we’re already applying the model to Germany. We rely on the Robert Koch Institute’s data for the analysis of the German figures, since the model takes into account how many test are conducted per day and per region. This is an important factor, since the total number of tests performed also has an impact on how many positive cases are found. However, the actual R value is independent of the kinds of fluctuations in testing that we observe at the weekend, for example.

rt-liveDaily forecast of the R value for German federal states can be found on the website.
Copyright: Forschungszentrum Jülich

What are the prospects for a German “” website?

Laura Helleckes: We learned a lot from the project and are currently examining in what form the model could be used to cope with a second wave in Germany. In that spirit, we are now making our daily updated forecasts available to the public at

Where do you still see problems?

Michael Osthege: Unlike with the positive results, the German testing figures are only available after a delay of a few days or weeks. That is why we have expanded the model to include a forecast of the total testing figures. The forecast is shown in the detailed view. However, the further back the latest figures date, the less reliable they become.

Further information:

Institute of Bio- and Geosciences – Biotechnology (IBG-1)

Research at Forschungszentrum to overcome the coronavirus crisis

Laura Helleckes
Institute of Bio- and Geosciences – Biotechnology (IBG-1)
Tel.: 02461 61- 3962

Michael Osthege
Institute of Bio- and Geosciences – Biotechnology (IBG-1)
Tel.: 02461 61- 8646

Press contact:
Erhard Zeiss, press officer
Tel.: 02461 61-1841