Dr. Philipp Berens is a bioinformatician who carries out neuroscientific data analyses. He specialises in the retina and has been studying the different retinal cell types at the Bernstein Center for Computational Neuroscience at the University of Tübingen for around four years. Berens is specifically interested in the retinal code, i.e. the visual signals that are transferred from the eyes into the brain where they are processed into images. Berens obtains the experimental data for his analyses from Prof. Dr. Thomas Euler and his team of researchers at the Institute for Ophthalmic Research at the Tübingen University Hospital. This cooperation was largely made possible by the Werner Reichardt Centre for Integrative Neurosciences, which is funded by the German Excellence Initiative.

The eye researchers record the activities of retinal nerve cells using two-photon microscopes and deliver the results to Berens and his team. "Our role is to process the data using statistical methods and make an objective evaluation," says Berens. "The quantity of data is so huge that it is impossible to do the analyses effectively by hand." In order to make the results transparent, Berens and his students develop data analysis programmes by adapting existing methods to their specific requirements.

Together with his experimental cooperation partners, Berens has spent the past few years concentrating on a specific class of retinal nerve cells. These cells are referred to as ganglion cells and their processes form the optic nerve that transfers signals from the eye into the brain. The starting point for Berens' research was the prevailing textbook opinion that only around 10 ganglion cell types existed. However, the researchers found over 30 anatomically different ones. "These newly discovered cell types fulfil different functions that were not known prior to our research," says Berens.

Berens and his colleagues have analysed the microscope images of more than 11,000 cells and created "fingerprints" for each ganglion cell type. "We found that there were a lot more ganglion cell types in existence than had previously been assumed; we were able to differentiate between at least 32 different types of responses," says Berens. The next objective is to characterise each cell type in greater detail. "One of our students is working on a rather special ganglion cell type that we did not originally plan to study. We will now take a closer look at these cells and examine how they respond to certain stimuli," says Berens.

Berens recently received the Bernstein Award for his work on retinal cells. The Bernstein Award is one of the most highly endowed scientific prizes for up-and-coming researchers in the world. The award is part of the Bernstein Network Computational Neuroscience which is a German Federal Ministry of Education and Research (BMBF) funding initiative that fosters computational neuroscience research in Germany. The award comes with a purse of up to 1.25 million euros and enables outstanding young researchers to establish an independent research group at a German institution. Over the next five years, Berens will use the money to expand his team at the University of Tübingen in order to analyse another retinal cell type, bipolar cells, in greater detail.

By focusing on bipolar cells, Berens is going one step further into the retina: this class of neurons is located between the photoreceptors and the ganglion cells, and transmits signals from photoreceptor cells to ganglion cells which Berens and his team have already studied. There are 14 different types of bipolar cells in the mouse retina. "The interesting thing about these cells is that the system itself is relatively simple, at least as far as input signals are concerned. The output signals and visual processes are extraordinarily complex. Berens will use computer models to understand the functional and biophysical properties of these cells by building on physiological and anatomical data from experimental research. Computer modelling allows him to find out whether these cells have different roles in visual processing.

These analyses will also be carried out in close cooperation with colleagues from the Department of Ophthalmology. Berens is careful to point out that the bioinformaticians and the biologists are equal partners. "It makes no sense at all to do the experiments and then analyse the data afterwards. So we are working together to design the experiments. It is a fairly iterative process in which we do experimental tests, analyse the data and then discuss step by step what needs to be changed in subsequent experiments. If the models produce different results from the measurements, we improve the models until the data match."

Although investigating retinal cells is basic research, Berens' work has obvious medical relevance. "If we know how these cells work, we can exploit this knowledge for therapeutic applications because retinal diseases are mainly degenerative processes. Take glaucoma for example. If we know which cells die first, we will understand this disease a lot better than we do now." Berens' research can also contribute to optimising restorative methods such as endoprosthetic ones. Berens explains how: "The problem with retinal chips is that the electrodes stimulate all retinal cells in the same way. The natural variety of responses cannot, therefore, be achieved. It goes without saying that we first need to find out what the response looks like."

In the medium to long term, the researchers want to create a complete virtual model of the retina so that they can carry out experiments with electrical stimulation in silico. This could also reduce the number of animal experiments. Ideally, the anatomically accurate model will react like real cells and contain more than just the "fingerprints" of the retinal cell types used. "To do this, we plan to develop a model to predict responses to simple light stimuli," says Berens. "Then we plan to establish a model based on anatomical data, which will be linked in a final step with disease models, for degenerative diseases in particular."