Search

Protein dynamics has been investigated on a wide range of time scales. Nano- and picosecond dynamics have been assigned to local fluctuations, while slower dynamics have been attributed to larger conformational changes. However, it is largely unknown how local fluctuations can lead to global allosteric changes. Here we show that molecule-spanning dynamics on the 100 ns time scale precede larger allosteric changes. We assign global real-space movements to dynamic modes on the 100 ns time scales, which became possible by a combination of single-molecule fluorescence, quasi-elastic neutron scattering and all-atom MD simulations. Additionally, we demonstrate the effect of Sba1, a co-chaperone of Hsp90, on these molecule-spanning dynamics, which implies functional importance of such dynamics. Our integrative approach provides comprehensive insights into molecule-spanning dynamics on the nanosecond time scale for a multi-domain protein and indicates that such dynamics are the molecular basis for allostery and large conformational changes in proteins., 29 main text pages, 7 main text figures; 27 SI pages, 15 SI figures

This work introduces a new proposal-free instance segmentation method that builds on single-instance segmentation masks predicted across the entire image in a sliding window style. In contrast to related approaches, our method concurrently predicts all masks, one for each pixel, and thus resolves any conflict jointly across the entire image. Specifically, predictions from overlapping masks are combined into edge weights of a signed graph that is subsequently partitioned to obtain all final instances concurrently. The result is a parameter-free method that is strongly robust to noise and prioritizes predictions with the highest consensus across overlapping masks. All masks are decoded from a low dimensional latent representation, which results in great memory savings strictly required for applications to large volumetric images. We test our method on the challenging CREMI 2016 neuron segmentation benchmark where it achieves competitive scores., Presented at GCPR 2020

Deep neural networks trained to inpaint partially occluded images show a deep understanding of image composition and have even been shown to remove objects from images convincingly. In this work, we investigate how this implicit knowledge of image composition can be leveraged for fully self-supervised instance separation. We propose a measure for the independence of two image regions given a fully self-supervised inpainting network and separate objects by maximizing this independence. We evaluate our method on two microscopy image datasets and show that it reaches similar segmentation performance to fully supervised methods.

Volumetric Electron Microscopy images can be used for connectomics, the study of brain connectivity at the cellular level. A prerequisite for this inquiry is the automatic identification of neural cells, which requires machine learning algorithms and in particular efficient image segmentation algorithms. In this thesis, we develop new algorithms for this task. In the first part we provide, for the first time in this field, a method for training a neural network to predict optimal input data for a watershed algorithm. We demonstrate its superior performance compared to other segmentation methods of its category. In the second part, we develop an efficient watershed-based algorithm for weighted graph partitioning, the \emph{Mutex Watershed}, which uses negative edge-weights for the first time. We show that it is intimately related to the multicut and has a cutting edge performance on a connectomics challenge. Our algorithm is currently used by the leaders of two connectomics challenges. Finally, motivated by inpainting neural networks, we create a method to learn the graph weights without any supervision.

Semantic instance segmentation is the task of simultaneously partitioning an image into distinct segments while associating each pixel with a class label. In commonly used pipelines, segmentation and label assignment are solved separately since joint optimization is computationally expensive. We propose a greedy algorithm for joint graph partitioning and labeling derived from the efficient Mutex Watershed partitioning algorithm. It optimizes an objective function closely related to the Symmetric Multiway Cut objective and empirically shows efficient scaling behavior. Due to the algorithm's efficiency it can operate directly on pixels without prior over-segmentation of the image into superpixels. We evaluate the performance on the Cityscapes dataset (2D urban scenes) and on a 3D microscopy volume. In urban scenes, the proposed algorithm combined with current deep neural networks outperforms the strong baseline of `Panoptic Feature Pyramid Networks' by Kirillov et al. (2019). In the 3D electron microscopy images, we show explicitly that our joint formulation outperforms a separate optimization of the partitioning and labeling problems.

We humans seem to have an innate understanding of the asymmetric progression of time, which we use to efficiently and safely perceive and manipulate our environment. Drawing inspiration from that, we address the problem of learning an arrow of time in a Markov (Decision) Process. We illustrate how a learned arrow of time can capture meaningful information about the environment, which in turn can be used to measure reachability, detect side-effects and to obtain an intrinsic reward signal. We show empirical results on a selection of discrete and continuous environments, and demonstrate for a class of stochastic processes that the learned arrow of time agrees reasonably well with a known notion of an arrow of time given by the celebrated Jordan-Kinderlehrer-Otto result., A shorter version of this work was presented at the Theoretical Phyiscs for Deep Learning Workshop, ICML 2019

The overhead of network communication creates a significant performance gap between local and distributed query processing in parallel main-memory database systems. We introduce a novel query engine designed to close this gap. It exploits locality in the data placement, leverages high-speed networks, and avoids the load imbalance caused by skew. A prototypical implementation for the high-performance in-memory database system HyPer outperforms competing systems in the renowned TPC-H benchmark. Teure Netzwerkkommunikation führt zu einer erheblichen Leistungslücke zwischen lokaler und verteilter Anfrageverarbeitung in parallelen Hauptspeicher-Datenbanksystemen. Wir präsentieren den Entwurf für eine neue verteilte Anfrageeinheit um diese Lücke zu schließen. Diese passt sich an die verfügbare Netzwerkbandbreite sowie unerwartete Eigenschaften der Daten an. Ein Prototyp für das hochperformante Hauptspeicher-Datenbanksystem HyPer schlägt konkurrierende Systeme im anerkannten TPC-H Benchmark.

Histamine is a neurotransmitter involved in a number of physiological and neuronal functions. In mammals, such as humans, and rodents, the histaminergic neurons found in the tuberomamillary nucleus project widely throughout the central nervous system. Histamine acts as positive modulator of \(GABA_{A}\) receptors \((GABA_{A}Rs)\) and, in high concentrations (10 mM), as negative modulator of the strychnine-sensitive glycine receptor. However, the exact molecular mechanisms by which histamine acts on \(GABA_{A}Rs\) are unknown. In our study, we aimed to identify amino acids potentially involved in the modulatory effect of histamine on \(GABA_{A}Rs\). We expressed \(GABA_{A}Rs\) with 12 different point mutations in Xenopus laevis oocytes and characterized the effect of histamine on GABA-induced currents using the two-electrode voltage clamp technique. Our data demonstrate that the amino acid residues \(\beta\)2(N265) and \(\beta\)2(M286), which are important for modulation by propofol, are not involved in the action of histamine. However, we found that histamine modulation is dependent on the amino acid residues \(\alpha\)1(R120), \(\beta\)2(Y157), \(\beta\)2(D163), \(\beta\)3(V175), and \(\beta\)3(Q185). We showed that the amino acid residues \(\beta\)2(Y157) and \(\beta\)3(Q185) mediate the positive modulatory effect of histamine on GABA-induced currents, whereas \(\alpha\)1(R120) and \(\beta\)2(D163) form a potential histamine interaction site in \(GABA_{A}Rs\).

G protein-coupled receptors (GPCRs) form the largest family of membrane receptors in the human genome. Advances in membrane protein crystallization so far resulted in the determination of 24 receptors available as high-resolution atomic structures. We performed the first phylogenetic analysis of GPCRs based on the available set of GPCR structures. We present a new phylogenetic tree of known human rhodopsin-like GPCR sequences based on this structure set. We can distinguish the three separate classes of small-ligand binding GPCRs, peptide binding GPCRs, and olfactory receptors. Analyzing different structural subdomains, we found that small molecule binding receptors most likely have evolved from peptide receptor precursors, with a rhodopsin/S1PR1 ancestor, most likely an ancestral opsin, forming the link between both classes. A light-activated receptor therefore seems to be the origin of the small molecule hormone receptors of the central nervous system. We find hints for a common evolutionary path of both ligand binding site and central sodium/water binding site. Surprisingly, opioid receptors exhibit both a binding cavity and a central sodium/water binding site similar to the one of biogenic amine receptors instead of peptide receptors, making them seemingly prone to bind small molecule ligands, e.g. opiates. Our results give new insights into the relationship and the pharmacological properties of rhodopsin-like GPCRs.