Forschungs- & Entwicklungsinformationsdienst der Gemeinschaft - CORDIS


COMPSELF Berichtzusammenfassung

Project ID: 267369
Gefördert unter: FP7-IDEAS-ERC
Land: United Kingdom

Final Report Summary - COMPSELF (Self-Organisation: From Molecules to Matter)

This project aimed to gain new insight into self-organising systems and exploit this knowledge throughout molecular science. We have been able to make progress in a wide variety of areas, including the development of new theory and simulation tools.
Novel theory and algorithms have been described for structure prediction, thermodynamic sampling, and rare events. For structure prediction new basin-hopping global optimisation methods have been introduced based on the principle of maximum symmetry. Schemes that employ free energy directly can locate the free energy global minimum faster than post-processing local minima, which grand and semi-grand basin-hopping approaches provide surveys of global minima as a function of size and composition. The problem of chemical ordering in alloys has been addressed using an approach based on multiminima.

Coarse-grained models have been developed using a general angle-axis coordinate system.
This approach can treat decorated rigid bodies, and has been enhanced using new move sets designed to prevent structural overlap.
A new local rigidification scheme permits the grouping of arbitrary sets of atoms into local rigid bodies, producing significant gains in efficiency by reducing the search space, and eliminating intra-rigid-body terms from the energy evaluation.

The pathways involved in large-scale conformational changes of biomolecules such as proteins and nucleic acids can involve many elementary steps. To treat the rare event kinetics associated with these events we have developed new interpolation schemes to identify initial paths, along with novel tools for calculating the associated rates.
In particular, the quasi-continuous interpolation scheme prevents chain-crossing in proteins and nucleic acids, providing kinetically relevant starting points. Methods based on population relaxation rather than reactive flux provide more accurate rates when spatial diffusion is important.

Energy landscapes where there are competing morphologies separated by high barriers pose particular problems for equilibrium sampling.
The energy landscape approach has been used to develop new tools for addressing this broken ergodicity, using formulations based on the contributions of local minima.
The same framework was used to facilitate the calculation of ligand binding free energies. These methods have been applied in combination with existing enhanced sampling procedures, including nested sampling and parallel tempering. Alternatively, an approximate scheme has been described, namely basin-sampling, which provides far greater efficiency for atomic clusters that feature multi-funnel landscapes.

These methods for structure prediction, thermodynamics and kinetics, have all been applied to propose design schemes for self-organising systems, ranging from fully atomisitic to coarse-grained models. A minimal patchy building block has been developed, which self-assembles into a Bernal spiral, and we have identified a family of left-handed building blocks, where the internal geometry can be tuned to favour either left- or right-handed helices.
An internally frustrated building block has been designed to model the formation of amyloid fibre morphologies.
Here the design principles can be extended to the macroscopic length scale, to explain the structure of certain seed pods, and helices formed from ellipsoidal magnets.

All the new methodology and potentials have been made available in public domain computer programs. Many new applications and collaborations have been reported and more are in progress.
For example, pathways and mechanisms for proton transfer to small molecules that are present in significantly different concentrations in tumours will assist in the development of new medical imaging tools. This work is being carried out in collaboration with the German Cancer Research Centre in Heidelberg and the Free University of Berlin.

Perhaps the greatest success involves the prediction of structure and pathways for nucleic acids.
The time scales involved make conventional simulations prohibitively expensive, but for tools based on geometry optimisation the pathways can be relatively straightforward. Recent applications have treated RNA repeats associated with human disease, along with pathways that connect different functional morphologies for RNA. Here we have characterised multi-funnel landscapes, which we suggest are associated with multi-functional behaviour, based on results for an intrinsically disordered protein, which provided a testing ground for many of the new methods we have developed.

Reported by

United Kingdom
Folgen Sie uns auf: RSS Facebook Twitter YouTube Verwaltet vom Amt für Veröffentlichungen der EU Nach oben