Expanding the toolbox for neutron and X-ray experiments

The data generated from experiments at large scale research infrastructures is often vast and very complex to analyse and interpret. Meanwhile, it is becoming increasingly important to find out more about protein behavior – to develop new types of food, more effective medicines, and more knowledge about the human body.

Mikael Lund, professor in theoretical chemistry at Lund University, is leading a working group research programme at LINXS concerned with how to make data analysis both easier and more effective. The aim is to build computer simulations and models of protein experiments to help researchers analyse and predict protein behavior. In October 2019, the programme had a tree day workshop on the topic of scattering in anisotropic systems, where participants identified scattering in concentrated protein solutions as an area to further develop.

– Different types of computations can help researchers understand and work with their data in new ways. Ultimately, it is about making research at large infrastructure more insightful and efficient, says Mikael Lund.

He explains that researchers are faced with the twofold problem of length and time scales when they conduct scattering experiments on protein solutions.

– Today, many experiments are limited in terms of spatial and temporal resolution. Computer simulations can complement these with a broader set of time and length scales and become part of a loop to extract as much information as possible from a scattering experiments.

Interaction Models and Analysis Tools

The programme’s work is split into two parts: one part is focused on creating interaction models for proteins; the other on creating an analysis tool to link computer simulations to experimental data.

To develop their models and simulations, Mikael Lund and his colleagues will need input from experiments conducted at large scale research infrastructures. They will therefore work together with another working group at LINXS, namely Antibodies in Solutions, and use their protein samples.

– Using a large set of well-defined samples is key for training our computer code and create descriptive models of the experiments. It will help us make predictions and analyse the experiments since the samples present a broad set of data, conducted under a well-defined set of sample conditions.

So far, Mikael Lund and his group have created a prototype model and a simulation tool for predictive scattering experiments. However, some of the analysis is still very demanding, even for the computer simulations.

– A major challenge is to create tools that can be used more broadly by researchers. You need tools that are both efficient and user friendly. One problem we need to solve is how to handle large data sets. In our simulations, we want to look at many interacting proteins as found in concentrated samples. Before, it was more common to only analyse one, perhaps two, proteins at a time.

The long-term ambition of the research programme is to include the computational tools in already established software, and make them widely available. This is important because timeslots at large scale research infrastructures are often limited and could be used more effectively. An accurate computational toolbox could guide researchers in the experimental design and make precious beam time much more efficient.

The research programme: Simulation, theory, and software development for anisotropic systems, falls under the theme Dynamics.