Rapid interpretation of small-angle X-ray scattering data.

The fundamental aim of structural analyses in biophysics is to reveal a mutual relation between a molecule's dynamic structure and its physiological function. Small-angle X-ray scattering (SAXS) is an experimental technique for structural characterization of macromolecules in solution and enables time-resolved analysis of conformational changes under physiological conditions. As such experiments measure spatially averaged low-resolution scattering intensities only, the sparse information obtained is not sufficient to uniquely reconstruct a three-dimensional atomistic model. Here, we integrate the information from SAXS into molecular dynamics simulations using computationally efficient native structure-based models. Dynamically fitting an initial structure towards a scattering intensity, such simulations produce atomistic models in agreement with the target data. In this way, SAXS data can be rapidly interpreted while retaining physico-chemical knowledge and sampling power of the underlying force field. We demonstrate our method's performance using the example of three protein systems. Simulations are faster than full molecular dynamics approaches by more than two orders of magnitude and consistently achieve comparable accuracy. Computational demands are reduced sufficiently to run the simulations on commodity desktop computers instead of high-performance computing systems. These results underline that scattering-guided structure-based simulations provide a suitable framework for rapid early-stage refinement of structures towards SAXS data with particular focus on minimal computational resources and time.

RNA contact prediction by data efficient deep learning.

On the path to full understanding of the structure-function relationship or even design of RNA, structure prediction would offer an intriguing complement to experimental efforts. Any deep learning on RNA structure, however, is hampered by the sparsity of labeled training data. Utilizing the limited data available, we here focus on predicting spatial adjacencies ("contact maps") as a proxy for 3D structure. Our model, BARNACLE, combines the utilization of unlabeled data through self-supervised pre-training and efficient use of the sparse labeled data through an XGBoost classifier. BARNACLE shows a considerable improvement over both the established classical baseline and a deep neural network. In order to demonstrate that our approach can be applied to tasks with similar data constraints, we show that our findings generalize to the related setting of accessible surface area prediction.

The Trimeric Major Capsid Protein of Mavirus is stabilized by its Interlocked N-termini Enabling Core Flexibility for Capsid Assembly.

Icosahedral viral capsids assemble with high fidelity from a large number of identical buildings blocks. The mechanisms that enable individual capsid proteins to form stable oligomeric units (capsomers) while affording structural adaptability required for further assembly into capsids are mostly unknown. Understanding these mechanisms requires knowledge of the capsomers' dynamics, especially for viruses where no additional helper proteins are needed during capsid assembly like for the Mavirus virophage that despite its complexity (triangulation number T = 27) can assemble from its major capsid protein (MCP) alone. This protein forms the basic building block of the capsid namely a trimer (MCP3) of double-jelly roll protomers with highly intertwined N-terminal arms of each protomer wrapping around the other two at the base of the capsomer, secured by a clasp that is formed by part of the C-terminus. Probing the dynamics of the capsomer with HDX mass spectrometry we observed differences in conformational flexibility between functional elements of the MCP trimer. While the N-terminal arm and clasp regions show above average deuterium incorporation, the two jelly-roll units in each protomer also differ in their structural plasticity, which might be needed for efficient assembly. Assessing the role of the N-terminal arm in maintaining capsomer stability showed that its detachment is required for capsomer dissociation, constituting a barrier towards capsomer monomerisation. Surprisingly, capsomer dissociation was irreversible since it was followed by a global structural rearrangement of the protomers as indicated by computational studies showing a rearrangement of the N-terminus blocking part of the capsomer forming interface.

Including residual contact information into replica-exchange MD simulations significantly enriches native-like conformations.

Proteins are complex biomolecules which perform critical tasks in living organisms. Knowledge of a protein's structure is essential for understanding its physiological function in detail. Despite the incredible progress in experimental techniques, protein structure determination is still expensive, time-consuming, and arduous. That is why computer simulations are often used to complement or interpret experimental data. Here, we explore how in silico protein structure determination based on replica-exchange molecular dynamics (REMD) can benefit from including contact information derived from theoretical and experimental sources, such as direct coupling analysis or NMR spectroscopy. To reflect the influence from erroneous and noisy data we probe how false-positive contacts influence the simulated ensemble. Specifically, we integrate varying numbers of randomly selected native and non-native contacts and explore how such a bias can guide simulations towards the native state. We investigate the number of contacts needed for a significant enrichment of native-like conformations and show the capabilities and limitations of this method. Adhering to a threshold of approximately 75% true-positive contacts within a simulation, we obtain an ensemble with native-like conformations of high quality. We find that contact-guided REMD is capable of delivering physically reasonable models of a protein's structure.