Back

Yksinkertaistamalla tarkempiin tuloksiin

Aplication Specialist Atte Sillanpää and Alex de Vries, a visiting lecturer at the Spring School in Computational Chemistry, discussed about coarse-graining. Photo: Heta Koski

More accurate results through simplification – computational methods in chemistry

Heta Koski

A computational chemist would say that there isn't a field of chemistry in which computational methods would not be of assistance. And they might well be right. Although the majority of contemporary chemists do not use computational methods as part of their routine research or daily work, the field is steadily growing. More and more chemists are using computer simulations to help them solve chemical problems, particularly in pharmaceutical design and materials research.

"Computational chemistry is in many ways cheaper, easier and faster than traditional empirical research," says CSC Application Specialist and chemist Atte Sillanpää.

"Simulations can also study things that would be impossible – or extremely difficult – to study empirically. Modelling is a fully-fledged research instrument, just as valid as, say, chromatography."

The fundamental theories of physics, such as quantum mechanics, very precisely describe what is happening in the chemical phenomena being researched. Therefore, although the behaviour and reactions of atoms and molecules can in principle be modelled mathematically, detailed quantum mechanical models are computationally extremely demanding. In order to use these theories to answer practically meaningful research hypotheses, they must be simplified, that is, make approximations that simplify the calculations. It is not possible to always calculate everything from the atomic level. In practice, researchers must accept a large number of approximations (or simplifications) on which the model is built.

Sillanpää has been involved in the organisation of six Spring Schools in Computational Chemistry, an annual event in Espoo that brings together top speakers and students from around the world.

"There's a lot happening in the field – that's why we started the school in the first place. Computational methods can be applied in most chemical research, even when you also engage in ‘wet' empirical chemistry," says Sillanpää.

No one is expected to become a top expert during the four-day course. Sillanpää says that this intensive, solution-based course is meant to immerse participants in a broad spectrum of methods that will open up a wave of opportunities and insights.


Magdalena Scharf and Matthäus Drabek took the course to obtain a general overview of computational methods. Photo: Atte Sillanpää

Magdalena Scharf and Matthäus Drabek from Marburg University in Germany have just began working on their doctoral theses. They took the course to obtain a general overview of computational methods.

"At this stage, I don't know which methods I'm going to need, so it's a good idea to learn as many as possible. It's also worthwhile learning what computational tools can be used for in general, and what software exists," says Scharf.

"I hope that at the end of the course the participants will have an understanding of everything that can be done with the aid of computational methods. The next time they come across something in their research that could benefit from computational methods, they can further develop their own computing competence or acquire a knowledgeable collaborative partner," says Sillanpää.

More tools in the chemist's bag

Although the majority of chemists do not routinely use computational methods, they are already established in the field. Modelling has been used in chemistry for decades and, for example, the software has become user friendly. A large proportion of CSC's computing resources are used either for research in materials science and chemistry or to study a variety of biosystems.

"Premier scientific publications value research that includes modelling," says Sillanpää.

The results obtained from modern measuring devices can be so difficult to understand that they should be repeated computationally. If the computational result does not match the measurement, it may be that the measurement has been incorrectly interpreted.

Modelling plays a major role in materials research. The EU-funded Novel Materials Discovery (NOMAD) Laboratory project, which was launched in 2015, is developing a new materials database that collates the millions of simulation results from all over the world performed over the years and converts them into a standardised format.

"The database contains simulation files that have been created using 40 of the most common software programmes, which means that the raw data is in 40 different formats. Without a standardised database structure, these results would not be easily reusable or utilisable in, for example, big data analyses," says Sillanpää

Using new tools also requires new skills. For instance, this year's Spring School introduced machine learning as one of the new methods available to chemists.

A better overview with coarse-graining

Coarse-graining is one of the commonly used methods in computational chemistry. It enables the simulations of larger model systems. In coarse-graining, molecules are not described in atomic scale detail. Instead, several atoms are combined to form an entity whose properties equate on average to those of the combined atoms.

In this video, Atte Sillanpää and Alex de Vries discuss the opportunities afforded by coarse-graining and what it can be used to study (the article continues after the video):


Camera: Heta Koski. Editing: Atte Sillanpää

"You could compare this to a computer game in which a wall looks like a smooth surface, even though in reality it would be composed of bricks," says Alex de Vries, a researcher at the University of Groningen and one of the Spring School's visiting speakers.

"By using coarse-graining methods, you can study molecular behaviour further than you could using atomistic modelling with the same resources. It enables you to obtain more reliable samples of all of the system's probable states, and to get a better overview of complex phenomena," says de Vries.

Coarse-grained modelling is not suitable for all phenomena, as a number of interactions are averaged. Suitable targets include cell membranes, as their properties are primarily defined by the interactions of large groups of molecules. Cell membrane model systems are also vast, and need to be simulated for a long time to describe their phenomena.

"You can return to a more precise atomic model from the structure generated by a coarse-grained model. You can then be sure that the structure is also correct at a detailed level, and avoid any artefacts arising from coarse-graining," says De Vries.

Facts:

  • The Spring School is one of several events that CSC arranges annually as part of the PRACE Advanced Training Center's (PATC) calendar. This first-class training event is free to participants.
  • The Spring School was held for the sixth time in 2017. There were participants of eleven different nationalities from six different countries.
  • The course's exercises and speakers' presentations are openly accessible on the School's website: https://events.prace-ri.eu/event/560


 



comments powered by Disqus